From 8d28fd9f9e85f5e0715b3be2337b7843fc4c943f Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Wed, 22 Jul 2026 16:54:31 -0500 Subject: results: close stability-margin training screen --- .../kp_innovation_stability_grid/margin_0p001.json | 33199 +++++++++++++++++++ .../kp_innovation_stability_grid/margin_0p003.json | 33151 ++++++++++++++++++ .../kp_innovation_stability_grid/margin_0p01.json | 32986 ++++++++++++++++++ .../kp_innovation_stability_grid/margin_0p03.json | 32986 ++++++++++++++++++ 4 files changed, 132322 insertions(+) create mode 100644 results/kp_innovation_stability_grid/margin_0p001.json create mode 100644 results/kp_innovation_stability_grid/margin_0p003.json create mode 100644 results/kp_innovation_stability_grid/margin_0p01.json create mode 100644 results/kp_innovation_stability_grid/margin_0p03.json (limited to 'results/kp_innovation_stability_grid') diff --git a/results/kp_innovation_stability_grid/margin_0p001.json b/results/kp_innovation_stability_grid/margin_0p001.json new file mode 100644 index 0000000..8f920e9 --- /dev/null +++ b/results/kp_innovation_stability_grid/margin_0p001.json @@ -0,0 +1,33199 @@ +{ + "first_any_nonfinite_groups": { + "bn_running_var": [ + 15 + ] + }, + "first_any_nonfinite_step": 252, + "first_nonfinite_by_group": { + "bn_running_var": { + "indices": [ + 15 + ], + "step": 252 + } + }, + "first_nonfinite_by_metric": {}, + "first_nonfinite_metrics": [], + "first_training_failure_groups": {}, + "first_training_failure_step": null, + "initial_parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4011031687259674, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3956821858882904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1162850856781006, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 0.0011125205783173442, + "max_applied_stability_margin": 0.001112517318688333, + "max_positive_residual_soma_slope": 0.0, + "min_residual_soma_slope": -0.0011125205783173442, + "mse": 8.080041138584974e-07, + "observations": 64, + "residual_traffic_rms_ratio": 0.05331967893579812, + "stability_margin": 0.001 + }, + "first_mse": 8.080041138584974e-07, + "last_mse": 8.080041138584974e-07, + "post_warmup_traffic_residual_rms_ratio": 0.05331967893579812, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_mixed_traffic_nonfinite_diagnosis_v1", + "provenance": { + "git_commit": "0ae690e7b3144da2c5b3384cd7169e86529fd26f", + "git_tracked_dirty": false + }, + "rule": "innovation", + "scope": "training_only_no_validation_or_test_evaluation", + "split": { + "dataset": "cifar10", + "input_layout": "NCHW", + "input_shape": [ + 3, + 32, + 32 + ], + "loader_seed": 0, + "normalization_mean": [ + 0.49140000343322754, + 0.4821999967098236, + 0.4465000033378601 + ], + "normalization_std": [ + 0.24699999392032623, + 0.2434999942779541, + 0.26159998774528503 + ], + "split_from_training_only": true, + "split_seed": 2027, + "test_examples": 10000, + "train_examples": 45000, + "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5", + "validation_class_counts": { + "0": 500, + "1": 500, + "2": 500, + "3": 500, + "4": 500, + "5": 500, + "6": 500, + "7": 500, + "8": 500, + "9": 500 + }, + "validation_examples": 5000, + "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b" + }, + "stability_margin": 0.001, + "test_evaluations": 0, + "traffic_calibration": { + "instruction_rms": [ + 0.008471464738249779, + 0.006617916747927666, + 0.003988116513937712, + 0.0043865954503417015, + 0.002713605063036084, + 0.003158062230795622, + 0.002004207344725728, + 0.003757704282179475, + 0.0027085733599960804, + 0.0030200635083019733, + 0.002056022873148322, + 0.002389939734712243, + 0.0016842411132529378, + 0.0034010226372629404, + 0.002549725119024515, + 0.0030542907770723104, + 0.0020918569061905146, + 0.002605808200314641, + 0.0018916188273578882 + ], + "realized_traffic_instruction_rms_ratio": [ + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 3.999999523162842, + 4.0, + 4.000000476837158, + 3.999999761581421, + 4.0, + 4.0, + 3.999999761581421, + 4.0, + 4.000000476837158, + 4.000000476837158, + 4.0, + 4.0, + 4.0 + ], + "target_ratio": 4.0, + "traffic_gain": [ + 0.0443243607878685, + 0.035817068070173264, + 0.013778294436633587, + 0.022762222215533257, + 0.007214079145342112, + 0.017218582332134247, + 0.004616298712790012, + 0.02042512409389019, + 0.007327491883188486, + 0.01611751690506935, + 0.004703814629465342, + 0.01256294921040535, + 0.003324519144371152, + 0.01805141195654869, + 0.006063441745936871, + 0.016296451911330223, + 0.0044285994954407215, + 0.013874167576432228, + 0.0036007820162922144 + ], + "unscaled_traffic_rms": [ + 0.7644973993301392, + 0.7390796542167664, + 1.157796859741211, + 0.7708553671836853, + 1.504616141319275, + 0.7336404919624329, + 1.736635684967041, + 0.7358984351158142, + 1.478581428527832, + 0.7495108246803284, + 1.7483876943588257, + 0.7609485983848572, + 2.0264477729797363, + 0.7536302804946899, + 1.6820316314697266, + 0.7496824264526367, + 1.8894071578979492, + 0.7512690424919128, + 2.10134220123291 + ] + }, + "trajectory": [ + { + "batch_loss": 2.9134020805358887, + "innovation_rms": 0.004314280146692243, + "instruction_rms": 0.004215397890235521, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40285658836364746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9875131845474243, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9400058388710022, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4058851897716522, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6322960257530212, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.040462132543325424, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40462133288383484, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.002788782119751, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9552946090698242, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9400270581245422, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3794170618057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.632293164730072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1172850131988525, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017318262558628915, + "step": 1, + "teaching_rms": 0.004314280146692243, + "traffic_rms": 0.01680093259933294 + }, + { + "batch_loss": 2.404780149459839, + "innovation_rms": 0.008364805764948397, + "instruction_rms": 0.00833062626564049, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7412737011909485, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0763678550720215, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9328936338424683, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40892329812049866, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4379763603210449, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09316778928041458, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5619637966156006, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9993336796760559, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4208698272705078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9328652024269104, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3803478181362152, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4379708766937256, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1185240745544434, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017541022660716856, + "step": 2, + "teaching_rms": 0.008364805764948397, + "traffic_rms": 0.015435872868067685 + }, + { + "batch_loss": 2.6168172359466553, + "innovation_rms": 0.010784622680986594, + "instruction_rms": 0.010762725939717065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.000083565711975, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2994701862335205, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8547738790512085, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4132925868034363, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24655263125896454, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1543719619512558, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6472303867340088, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9965748190879822, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5121314525604248, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8547332882881165, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38735532760620117, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24654479324817657, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.118544101715088, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017465313344570674, + "step": 3, + "teaching_rms": 0.010784622680986594, + "traffic_rms": 0.013761923805702077 + }, + { + "batch_loss": 2.7420530319213867, + "innovation_rms": 0.015101907322761529, + "instruction_rms": 0.01508791227807707, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1990811824798584, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.632760763168335, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7800410389900208, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41377636790275574, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13150231540203094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21246710419654846, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7430827021598816, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9923238158226013, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5166279077529907, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7799895405769348, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3824352025985718, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1314937174320221, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1188156604766846, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019142024127077096, + "step": 4, + "teaching_rms": 0.015101907322761529, + "traffic_rms": 0.011790454467010523 + }, + { + "batch_loss": 2.6681129932403564, + "innovation_rms": 0.021745348459835948, + "instruction_rms": 0.021736191174377632, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.366495966911316, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.937692165374756, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7051478028297424, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40896308422088623, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2420644909143448, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2785335183143616, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.908073365688324, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9864872694015503, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9225757122039795, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7050864696502686, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.389405220746994, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24210527539253235, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.120052456855774, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.023858995337201278, + "step": 5, + "teaching_rms": 0.021745348459835948, + "traffic_rms": 0.009861686838646088 + }, + { + "batch_loss": 2.5410773754119873, + "innovation_rms": 0.03161050584076436, + "instruction_rms": 0.03160478552675137, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.497052788734436, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1736562252044678, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6417884826660156, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4034843146800995, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3599436581134796, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35688528418540955, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9045425653457642, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9801166653633118, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.632317066192627, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6417182683944702, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40182235836982727, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35999035835266113, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1228549480438232, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03262523738000186, + "step": 6, + "teaching_rms": 0.03161050584076436, + "traffic_rms": 0.00808130511414367 + }, + { + "batch_loss": 2.4351558685302734, + "innovation_rms": 0.043569255684200156, + "instruction_rms": 0.04356503256822105, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6036124229431152, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.341135263442993, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5801504254341125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40055471658706665, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.431743323802948, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43270638585090637, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1720218658447266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9740140438079834, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6706738471984863, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.580072283744812, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40582361817359924, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4317953586578369, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.125178337097168, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.044085947902322724, + "step": 7, + "teaching_rms": 0.043569255684200156, + "traffic_rms": 0.006772799552875853 + }, + { + "batch_loss": 2.5690979957580566, + "innovation_rms": 0.06277274229419771, + "instruction_rms": 0.0627697821179615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6933177709579468, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5890214443206787, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5379677414894104, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40465036034584045, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2813761234283447, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5083223581314087, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6812043190002441, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9675063490867615, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.631951093673706, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5380546450614929, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40134167671203613, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28143295645713806, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1295965909957886, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06304237943594226, + "step": 8, + "teaching_rms": 0.06277274229419771, + "traffic_rms": 0.005855163925605933 + }, + { + "batch_loss": 2.625016212463379, + "innovation_rms": 0.07719772001950945, + "instruction_rms": 0.07719563540338739, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7805299758911133, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.895613431930542, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6706415414810181, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41277751326560974, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32049500942230225, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6078395843505859, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.054093599319458, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9589334726333618, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9417738914489746, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6707417368888855, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39528000354766846, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32047733664512634, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1339977979660034, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.07739447296859367, + "step": 9, + "teaching_rms": 0.07719772001950945, + "traffic_rms": 0.005539915514114038 + }, + { + "batch_loss": 2.385200023651123, + "innovation_rms": 0.05892665631289755, + "instruction_rms": 0.05892394091216684, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8802694082260132, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.199463844299316, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5029260516166687, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42270928621292114, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44575291872024536, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.706757128238678, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1275370121002197, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9545166492462158, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4545899629592896, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7849694490432739, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3909294903278351, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4457341134548187, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1379475593566895, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.05921746771509114, + "step": 10, + "teaching_rms": 0.05892665631289755, + "traffic_rms": 0.005907124308932213 + }, + { + "batch_loss": 2.3858439922332764, + "innovation_rms": 0.05922973322093472, + "instruction_rms": 0.059227262345814914, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.000638246536255, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.539255142211914, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45283082127571106, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42732590436935425, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.383269727230072, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8336042165756226, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.028526544570923, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9502367377281189, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.487173318862915, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8218613862991333, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38911938667297363, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3832499086856842, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1414909362792969, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.059568635358929756, + "step": 11, + "teaching_rms": 0.05922973322093472, + "traffic_rms": 0.006371710727349443 + }, + { + "batch_loss": 2.468273162841797, + "innovation_rms": 0.03542001657792386, + "instruction_rms": 0.03541633562789737, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1579694747924805, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.980679988861084, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7554428577423096, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4265801012516022, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19413445889949799, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9429590106010437, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9570932388305664, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.052476167678833, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1782177686691284, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1116492748260498, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3921002745628357, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19430381059646606, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1446685791015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03608006794061373, + "step": 12, + "teaching_rms": 0.03542001657792386, + "traffic_rms": 0.006887500373184148 + }, + { + "batch_loss": 2.6506686210632324, + "innovation_rms": 0.018788653669476334, + "instruction_rms": 0.018780233236093458, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.349302053451538, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4652533531188965, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6728109121322632, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42706456780433655, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3134482502937317, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8546806573867798, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0081267356872559, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5277886390686035, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0960273742675781, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7406504154205322, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7244151830673218, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3960869014263153, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3134516477584839, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.147286057472229, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02025669419623844, + "step": 13, + "teaching_rms": 0.018788653669476334, + "traffic_rms": 0.007595926514385619 + }, + { + "batch_loss": 2.6818325519561768, + "innovation_rms": 0.013410348346690427, + "instruction_rms": 0.013396710863235812, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8911752700805664, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.963738918304443, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6065265536308289, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42834344506263733, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45273175835609436, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8809733390808105, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1049840450286865, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2810685634613037, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0975100994110107, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7025794982910156, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3506886959075928, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3980030417442322, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45284542441368103, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1495933532714844, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.015841423748191957, + "step": 14, + "teaching_rms": 0.013410348346690427, + "traffic_rms": 0.008454007335770985 + }, + { + "batch_loss": 2.527742385864258, + "innovation_rms": 0.007363051569777398, + "instruction_rms": 0.007338701816055204, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7157533168792725, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.491358280181885, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5475509762763977, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4312750995159149, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6845588088035583, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9046279191970825, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1237852573394775, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7893476486206055, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0696816444396973, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.155160903930664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2230571508407593, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.397286981344223, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6845816969871521, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1611382961273193, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.011624141638037768, + "step": 15, + "teaching_rms": 0.007363051569777398, + "traffic_rms": 0.009010641555247294 + }, + { + "batch_loss": 2.3838140964508057, + "innovation_rms": 0.005181190616183203, + "instruction_rms": 0.00514848803240417, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.561798095703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.048993110656738, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4914935827255249, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4324367046356201, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5842943787574768, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9259080290794373, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1852543354034424, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.385547161102295, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0113766193389893, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1953179836273193, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1317695379257202, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3993077278137207, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5843178033828735, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.180955410003662, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010587993504240794, + "step": 16, + "teaching_rms": 0.005181190616183203, + "traffic_rms": 0.009251594637489627 + }, + { + "batch_loss": 2.547581911087036, + "innovation_rms": 0.0039365934072199986, + "instruction_rms": 0.003898412575118581, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.358227252960205, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.893908977508545, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4444495141506195, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4336443543434143, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27298682928085327, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.945050835609436, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2432277202606201, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7217235565185547, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9451903104782104, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8906644582748413, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0822420120239258, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40357503294944763, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27301886677742004, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.198695182800293, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009979296015382723, + "step": 17, + "teaching_rms": 0.0039365934072199986, + "traffic_rms": 0.009191226781393698 + }, + { + "batch_loss": 2.3952927589416504, + "innovation_rms": 0.0035309734968218404, + "instruction_rms": 0.0034913901355046713, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.04785680770874, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.110018730163574, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.402903288602829, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.435086727142334, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4615247845649719, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9622699022293091, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2562741041183472, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6566531658172607, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0059195756912231, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6863683462142944, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9881696105003357, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4044482111930847, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4612559676170349, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.214482307434082, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009587020824190887, + "step": 18, + "teaching_rms": 0.0035309734968218404, + "traffic_rms": 0.008929872503036599 + }, + { + "batch_loss": 2.3457720279693604, + "innovation_rms": 0.00413936907524671, + "instruction_rms": 0.0041057837402386456, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.589974880218506, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.043855667114258, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3672158718109131, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43348565697669983, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43602538108825684, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9777573943138123, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2470358610153198, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1894986629486084, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0807547569274902, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6149739027023315, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8939145803451538, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4018407166004181, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4357518255710602, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.228546380996704, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009419454293751801, + "step": 19, + "teaching_rms": 0.00413936907524671, + "traffic_rms": 0.008474315090794698 + }, + { + "batch_loss": 2.4054815769195557, + "innovation_rms": 0.005709442725676036, + "instruction_rms": 0.005684006301395367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.961775302886963, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.527785301208496, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3349299132823944, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4328490197658539, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2834430932998657, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9916859865188599, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2772489786148071, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1773195266723633, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.133926510810852, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.484800934791565, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8030108213424683, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39626529812812805, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2833823561668396, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2412993907928467, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009793515724206016, + "step": 20, + "teaching_rms": 0.005709442725676036, + "traffic_rms": 0.007978086224157998 + }, + { + "batch_loss": 2.3858842849731445, + "innovation_rms": 0.00790424246175192, + "instruction_rms": 0.007883133561723227, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.168447971343994, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.045133590698242, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41760537028312683, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43652498722076416, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23315292596817017, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0042109489440918, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.320427417755127, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9641079902648926, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1738961935043335, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85701322555542, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7239866852760315, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38947051763534546, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2330913245677948, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2528718709945679, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010868637452550647, + "step": 21, + "teaching_rms": 0.00790424246175192, + "traffic_rms": 0.0074907983926515926 + }, + { + "batch_loss": 2.379261016845703, + "innovation_rms": 0.012757615891033651, + "instruction_rms": 0.012743572765799136, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.232514381408691, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.73179817199707, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43784299492836, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.439242422580719, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1804410219192505, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0154434442520142, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4920538663864136, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1031882762908936, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1954916715621948, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7835344076156616, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6550435423851013, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38557934761047363, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18037866055965424, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2632819414138794, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.014585301142244728, + "step": 22, + "teaching_rms": 0.012757615891033651, + "traffic_rms": 0.00710132021769147 + }, + { + "batch_loss": 2.3981664180755615, + "innovation_rms": 0.01946438172285764, + "instruction_rms": 0.01945447898377986, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.198212623596191, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.357210159301758, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38384032249450684, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44009876251220703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1289975345134735, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0253647565841675, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.643656849861145, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.962242841720581, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2793664932250977, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7107808589935303, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5865539908409119, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.386234849691391, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1288774013519287, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.27260160446167, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02062097867961428, + "step": 23, + "teaching_rms": 0.01946438172285764, + "traffic_rms": 0.006844002958631929 + }, + { + "batch_loss": 2.5222861766815186, + "innovation_rms": 0.010745208076393306, + "instruction_rms": 0.01072540692039987, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.096402168273926, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.56955909729004, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4159364104270935, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4395813047885895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25639355182647705, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.034304141998291, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7779971361160278, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4878625869750977, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4604828357696533, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8111637830734253, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5277941226959229, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3893847167491913, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25627225637435913, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2809491157531738, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.012626014122900625, + "step": 24, + "teaching_rms": 0.010745208076393306, + "traffic_rms": 0.00666883645046158 + }, + { + "batch_loss": 2.4369590282440186, + "innovation_rms": 0.0077751739396267885, + "instruction_rms": 0.007744900529103618, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.943174362182617, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.88022994995117, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4904269278049469, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4387415051460266, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3763539791107178, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0423325300216675, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8981735706329346, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.718435287475586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6948940753936768, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3441126346588135, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.490629643201828, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39337632060050964, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3762316405773163, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2884658575057983, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010178711828797619, + "step": 25, + "teaching_rms": 0.0077751739396267885, + "traffic_rms": 0.0066059106098271285 + }, + { + "batch_loss": 2.3414671421051025, + "innovation_rms": 0.005495049098587118, + "instruction_rms": 0.005448693178497744, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.755153656005859, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.4381103515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5105559825897217, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43784549832344055, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.452195942401886, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0495744943618774, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0063323974609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7196969985961914, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.00107479095459, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.061807870864868, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5107602477073669, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40096357464790344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4520726799964905, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2952179908752441, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.008639300720295605, + "step": 26, + "teaching_rms": 0.005495049098587118, + "traffic_rms": 0.006704773831609092 + }, + { + "batch_loss": 2.3898725509643555, + "innovation_rms": 0.005264387479162878, + "instruction_rms": 0.005211899684008178, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.541981220245361, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.30622482299805, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5295839905738831, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43734636902809143, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2991156578063965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0561070442199707, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.103675365447998, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9609988927841187, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.399129867553711, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9805500507354736, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5297896862030029, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40509018301963806, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29899153113365173, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3045707941055298, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.008546279755742738, + "step": 27, + "teaching_rms": 0.005264387479162878, + "traffic_rms": 0.006770492727304122 + }, + { + "batch_loss": 2.425920009613037, + "innovation_rms": 0.0054650598877894535, + "instruction_rms": 0.005404656008786717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.334561824798584, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.06987762451172, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4896738827228546, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4388366639614105, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21155643463134766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0618737936019897, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.191283941268921, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.078735828399658, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9212019443511963, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.22072172164917, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4898808002471924, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40673986077308655, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21166637539863586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3251409530639648, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.00870859530271533, + "step": 28, + "teaching_rms": 0.0054650598877894535, + "traffic_rms": 0.006827879342492447 + }, + { + "batch_loss": 2.3441388607025146, + "innovation_rms": 0.005535115088661288, + "instruction_rms": 0.005460429326542259, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.1619744300842285, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.1959342956543, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7736200094223022, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44154462218284607, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37411344051361084, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0668495893478394, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2701315879821777, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1178388595581055, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.599641799926758, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.784397602081299, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7737151384353638, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40898725390434265, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3742240071296692, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3436408042907715, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0087965159602869, + "step": 29, + "teaching_rms": 0.005535115088661288, + "traffic_rms": 0.006895752021871712 + }, + { + "batch_loss": 2.368084192276001, + "innovation_rms": 0.005389533284145447, + "instruction_rms": 0.0052902060590641065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.041954040527344, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.312015533447266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0376800298690796, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4468590021133423, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33109137415885925, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0713183879852295, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.341094493865967, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2454261779785156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4578680992126465, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.58226203918457, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0376139879226685, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40918418765068054, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3312024772167206, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3602752685546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.008840105151828337, + "step": 30, + "teaching_rms": 0.005389533284145447, + "traffic_rms": 0.0070944559659790645 + }, + { + "batch_loss": 2.5144031047821045, + "innovation_rms": 0.0052213643518774595, + "instruction_rms": 0.005156316726452567, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.0434722900390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.07014083862305, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3977138996124268, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45198774337768555, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3122447729110718, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0753440856933594, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.404961109161377, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3582231998443604, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.316100120544434, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.582319259643555, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3976799249649048, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3998730182647705, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31193551421165466, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3752409219741821, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.008795817454818711, + "step": 31, + "teaching_rms": 0.0052213643518774595, + "traffic_rms": 0.007133920252543875 + }, + { + "batch_loss": 2.686887741088867, + "innovation_rms": 0.00624827801313433, + "instruction_rms": 0.006109863596426024, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.141467094421387, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 60.939788818359375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6666364669799805, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4504149258136749, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3659907281398773, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0791798830032349, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4624412059783936, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3037805557250977, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.3981523513793945, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.820523262023926, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.666770577430725, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39566344022750854, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.365777850151062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.388698697090149, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.00951634635743088, + "step": 32, + "teaching_rms": 0.00624827801313433, + "traffic_rms": 0.007309486633154829 + }, + { + "batch_loss": 2.3210976123809814, + "innovation_rms": 0.011269880848873686, + "instruction_rms": 0.011162359834983112, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.242284774780273, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.50996398925781, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.514834403991699, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4486601650714874, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3146807849407196, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0827010869979858, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5141732692718506, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.181101083755493, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.76437520980835, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.662229537963867, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.514824390411377, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3919355571269989, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3144671618938446, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4008013010025024, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.013450483619036129, + "step": 33, + "teaching_rms": 0.011269880848873686, + "traffic_rms": 0.007509501503094817 + }, + { + "batch_loss": 2.5001437664031982, + "innovation_rms": 0.009120765570964669, + "instruction_rms": 0.00902819981650538, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.443989276885986, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.70592498779297, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.796175718307495, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4516103267669678, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3545142412185669, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.085100769996643, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.560732126235962, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.130371570587158, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.175087928771973, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.107124328613281, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7961766719818115, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3950727581977844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3547082543373108, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4116394519805908, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.011887685807617515, + "step": 34, + "teaching_rms": 0.009120765570964669, + "traffic_rms": 0.007750009434768784 + }, + { + "batch_loss": 2.5703585147857666, + "innovation_rms": 0.007164937950964555, + "instruction_rms": 0.00704312701095367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.61251449584961, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.6119613647461, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4761815071105957, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45862025022506714, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5830013155937195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3125272989273071, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.602635145187378, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1281239986419678, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.584784507751465, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.096961975097656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.476182222366333, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39541611075401306, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5829732418060303, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4223979711532593, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010683953864827279, + "step": 35, + "teaching_rms": 0.007164937950964555, + "traffic_rms": 0.008033164275922672 + }, + { + "batch_loss": 2.66990065574646, + "innovation_rms": 0.0067844862212737295, + "instruction_rms": 0.006618114668852697, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.354507446289062, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.8605499267578, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.483146905899048, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4569997489452362, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4691130518913269, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5378098487854004, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.640347719192505, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2185347080230713, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.014994621276855, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.302102088928223, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.483112335205078, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40034040808677673, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46912962198257446, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.536531925201416, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010620140198263722, + "step": 36, + "teaching_rms": 0.0067844862212737295, + "traffic_rms": 0.008304535076999886 + }, + { + "batch_loss": 2.9839792251586914, + "innovation_rms": 0.00840781459651273, + "instruction_rms": 0.008235440152153104, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.71832275390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 326.4217834472656, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5633766651153564, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4561593532562256, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9475187063217163, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7438963651657104, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6742889881134033, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.346338987350464, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.496774673461914, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.817800521850586, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5633420944213867, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41601139307022095, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9475021362304688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7426185607910156, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.011927012965876595, + "step": 37, + "teaching_rms": 0.00840781459651273, + "traffic_rms": 0.008643745342744586 + }, + { + "batch_loss": 2.7375218868255615, + "innovation_rms": 0.006555360830405271, + "instruction_rms": 0.00623967189976505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.55860900878906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 453.20263671875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.629735231399536, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46117904782295227, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34863799810409546, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.966345191001892, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.704836130142212, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6251702308654785, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.098482131958008, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.01707649230957, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6297004222869873, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4048788547515869, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34856313467025757, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.00178861618042, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010880609896956009, + "step": 38, + "teaching_rms": 0.006555360830405271, + "traffic_rms": 0.008936352722154138 + }, + { + "batch_loss": 2.623648166656494, + "innovation_rms": 0.006098271294076988, + "instruction_rms": 0.0055885279516034655, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.631011962890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 545.9656372070312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1843693256378174, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47056108713150024, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49921444058418274, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.223440408706665, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7323286533355713, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.947713851928711, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.924043655395508, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.25560760498047, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.184401035308838, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3889288008213043, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4994867146015167, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.258880376815796, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010763504142646506, + "step": 39, + "teaching_rms": 0.006098271294076988, + "traffic_rms": 0.009208795348209554 + }, + { + "batch_loss": 2.831696033477783, + "innovation_rms": 0.0070134600452572545, + "instruction_rms": 0.006277328921590296, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 45.722328186035156, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 584.3040771484375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0365447998046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46820181608200073, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4995177388191223, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4342548847198486, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.779611825942993, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.587331771850586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.174602508544922, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.505584716796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.036478042602539, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4115910530090332, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49994054436683655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.469691276550293, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.011418549981677984, + "step": 40, + "teaching_rms": 0.0070134600452572545, + "traffic_rms": 0.009547115030589899 + }, + { + "batch_loss": 2.474534749984741, + "innovation_rms": 0.0063479953264602, + "instruction_rms": 0.004750065205997741, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.59479522705078, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 569.25048828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.572122573852539, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4833105504512787, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6495941281318665, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7035975456237793, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.24845290184021, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.688411712646484, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.23966407775879, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.650619506835938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.57205581665039, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4552740454673767, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6495656967163086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6408660411834717, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01109655402069207, + "step": 41, + "teaching_rms": 0.0063479953264602, + "traffic_rms": 0.010029649008623446 + }, + { + "batch_loss": 2.754695177078247, + "innovation_rms": 0.005330297280913251, + "instruction_rms": 0.003948268226950786, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.355655670166016, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 543.6185913085938, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.920133113861084, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5467624068260193, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7517147064208984, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.245443820953369, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7670419216156006, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.185891151428223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.597272872924805, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.576087951660156, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.920133113861084, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5187286138534546, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7517102360725403, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2808735370635986, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01023445908153999, + "step": 42, + "teaching_rms": 0.005330297280913251, + "traffic_rms": 0.009436344488608302 + }, + { + "batch_loss": 2.4612503051757812, + "innovation_rms": 0.007130418146044051, + "instruction_rms": 0.004190120336568132, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.30167007446289, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 547.3668212890625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.086236953735352, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5605806112289429, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5322777628898621, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.788970470428467, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.432836055755615, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.6579413414001465, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.092668533325195, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.953948974609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.086374282836914, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5325495600700378, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5322732329368591, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8243966102600098, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.011534432869864704, + "step": 43, + "teaching_rms": 0.007130418146044051, + "traffic_rms": 0.010748938139340766 + }, + { + "batch_loss": 2.7207729816436768, + "innovation_rms": 0.009402034661352239, + "instruction_rms": 0.0060072260648614565, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.61182403564453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 565.27001953125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.469488143920898, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5407077074050903, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5891925096511841, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.373699188232422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.293247699737549, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.60411548614502, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.09623718261719, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60.035675048828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.46942138671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5126794576644897, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5894664525985718, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3823089599609375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.013523755982040704, + "step": 44, + "teaching_rms": 0.009402034661352239, + "traffic_rms": 0.012116108042731442 + }, + { + "batch_loss": 2.544060468673706, + "innovation_rms": 0.009478095499445552, + "instruction_rms": 0.0061882158624787815, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.92673110961914, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 579.1355590820312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.50889778137207, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5265941619873047, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5892898440361023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.10485315322876, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.299204349517822, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.059568405151367, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.1229248046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.2668685913086, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.508922576904297, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49856865406036377, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5896026492118835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.1402716636657715, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.013256601258062444, + "step": 45, + "teaching_rms": 0.009478095499445552, + "traffic_rms": 0.01173324463378018 + }, + { + "batch_loss": 2.5006136894226074, + "innovation_rms": 0.009677593001732264, + "instruction_rms": 0.0039450628645256065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.48056411743164, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 771.8171997070312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.204631805419922, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5365698337554932, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44392427802085876, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.831103801727295, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.487672805786133, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.884683609008789, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.36331558227539, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.40391540527344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.204657554626465, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5085471272468567, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4436165392398834, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.859162330627441, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01595303347402654, + "step": 46, + "teaching_rms": 0.009677593001732264, + "traffic_rms": 0.015460817428141706 + }, + { + "batch_loss": 2.6954877376556396, + "innovation_rms": 0.01194391173820931, + "instruction_rms": 0.004980979027351162, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.4835319519043, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1088.2783203125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.84088134765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5632648468017578, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6376677751541138, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.625353813171387, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.907270431518555, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.195976257324219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.201416015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.38101959228516, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.840970993041992, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5352449417114258, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6374536752700806, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.633961200714111, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01859979820913323, + "step": 47, + "teaching_rms": 0.01194391173820931, + "traffic_rms": 0.017927804157559023 + }, + { + "batch_loss": 2.3461050987243652, + "innovation_rms": 0.016856443826614137, + "instruction_rms": 0.00514150040561309, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.6319580078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1321.291259765625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.334758758544922, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5962855815887451, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4572826623916626, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.394407272338867, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.77993106842041, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.726608276367188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.78241729736328, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.8099822998047, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.334815979003906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5682684779167175, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45706844329833984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.403013706207275, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020810632038318396, + "step": 48, + "teaching_rms": 0.016856443826614137, + "traffic_rms": 0.020162434430758578 + }, + { + "batch_loss": 2.39064359664917, + "innovation_rms": 0.015937384397497346, + "instruction_rms": 0.006469466718200042, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.86246871948242, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1397.27734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.227352142333984, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6328738331794739, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5100312829017639, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.001917839050293, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.894990921020508, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.150598526000977, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.00334167480469, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 152.20925903320312, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.22744369506836, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6048595309257507, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5098205208778381, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.913662910461426, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021796726468786665, + "step": 49, + "teaching_rms": 0.015937384397497346, + "traffic_rms": 0.020809778740124414 + }, + { + "batch_loss": 2.2963860034942627, + "innovation_rms": 0.01880665178793239, + "instruction_rms": 0.0037999311678123377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.92341613769531, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2311.777587890625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.70451736450195, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6623086929321289, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5481275916099548, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.272369384765625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.374589920043945, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.79599380493164, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.67786407470703, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.74522399902344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.70460510253906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6342971920967102, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5479938387870789, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.184123039245605, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.028796734258958984, + "step": 50, + "teaching_rms": 0.01880665178793239, + "traffic_rms": 0.02854397230489354 + }, + { + "batch_loss": 2.444526433944702, + "innovation_rms": 0.021684461293692076, + "instruction_rms": 0.003476851259946417, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.16082763671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3971.3955078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.140560150146484, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6825862526893616, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3816671669483185, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.823841094970703, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.234642028808594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.60052490234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.8155746459961, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.3771209716797, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6545775532722473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38147562742233276, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.735603332519531, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.034864765018814636, + "step": 51, + "teaching_rms": 0.021684461293692076, + "traffic_rms": 0.03469377438877455 + }, + { + "batch_loss": 2.3215389251708984, + "innovation_rms": 0.03184894416615038, + "instruction_rms": 0.004255285322085491, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 133.83192443847656, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5166.6357421875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.52032470703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6993050575256348, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45591676235198975, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.274328231811523, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.875272750854492, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.406314849853516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.78555297851562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 259.69970703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.520172119140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.671299159526825, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4556786119937897, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.251049041748047, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04312167804328352, + "step": 52, + "teaching_rms": 0.03184894416615038, + "traffic_rms": 0.042911975579112735 + }, + { + "batch_loss": 2.2843565940856934, + "innovation_rms": 0.03500227466337404, + "instruction_rms": 0.005810912112417346, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 142.21612548828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5620.7744140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.99102783203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7102311253547668, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3496969938278198, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.613441467285156, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.35268211364746, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.774085998535156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.17391967773438, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 323.8836669921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.99076843261719, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6822280287742615, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3499116897583008, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.607460021972656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.043799765529406745, + "step": 53, + "teaching_rms": 0.03500227466337404, + "traffic_rms": 0.043413247552843606 + }, + { + "batch_loss": 2.321927070617676, + "innovation_rms": 0.04381786791567342, + "instruction_rms": 0.022583264951377423, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 133.6927032470703, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6222.9755859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.98165130615234, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7213448882102966, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3477972149848938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.38142967224121, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.601924896240234, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.492431640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 223.60299682617188, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 374.29071044921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.9815444946289, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6933445930480957, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34775587916374207, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.375450134277344, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0587362405910767, + "step": 54, + "teaching_rms": 0.04381786791567342, + "traffic_rms": 0.05421362798077179 + }, + { + "batch_loss": 2.371020793914795, + "innovation_rms": 0.05164854878533321, + "instruction_rms": 0.0024001368977691106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.9083251953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8092.984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.39114379882812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7345688343048096, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5053253173828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.389760971069336, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.04450988769531, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.42586517333984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 270.0369567871094, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 464.3396301269531, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.39130401611328, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7065713405609131, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5051104426383972, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.38378143310547, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06766853624694771, + "step": 55, + "teaching_rms": 0.05164854878533321, + "traffic_rms": 0.06762661369441911 + }, + { + "batch_loss": 2.3474984169006348, + "innovation_rms": 0.06831121059428605, + "instruction_rms": 0.0018550380340507954, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 174.78004455566406, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9636.9716796875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.18709564208984, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7482574582099915, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5804845690727234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.892513275146484, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.25690460205078, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.12393951416016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 330.57745361328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 605.4049072265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.18717956542969, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7202627658843994, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5802696347236633, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.90110778808594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0853451964211083, + "step": 56, + "teaching_rms": 0.06831121059428605, + "traffic_rms": 0.08532543836961197 + }, + { + "batch_loss": 2.288336753845215, + "innovation_rms": 0.07552554744922213, + "instruction_rms": 0.002403239901216775, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.25482177734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11443.7451171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.29129028320312, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.765254020690918, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4113655388355255, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.953243255615234, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.12307357788086, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.66168975830078, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 403.18621826171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 726.0875244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.2914581298828, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7372621297836304, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41115429997444153, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.79291534423828, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.09283737099739349, + "step": 57, + "teaching_rms": 0.07552554744922213, + "traffic_rms": 0.09280610797779268 + }, + { + "batch_loss": 2.4006497859954834, + "innovation_rms": 0.1006022151723078, + "instruction_rms": 0.002767534066079747, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 181.7641143798828, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20997.27734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.04879760742188, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7784620523452759, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40119972825050354, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.95812225341797, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.45753479003906, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.34458923339844, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 495.35968017578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 921.7347412109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.0489501953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7504729628562927, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40098845958709717, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.79781341552734, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.12657061978576484, + "step": 58, + "teaching_rms": 0.1006022151723078, + "traffic_rms": 0.1265400108666093 + }, + { + "batch_loss": 2.169710874557495, + "innovation_rms": 0.12908477576895058, + "instruction_rms": 0.0023492153151439053, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 179.33177185058594, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 36092.14453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.256591796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7854858040809631, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3160249888896942, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 94.32564544677734, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.01286315917969, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.55328369140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 613.12353515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1177.638427734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.256591796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7574995160102844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31626343727111816, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 94.16535186767578, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.16139479207864538, + "step": 59, + "teaching_rms": 0.12908477576895058, + "traffic_rms": 0.16137758865611718 + }, + { + "batch_loss": 2.384819984436035, + "innovation_rms": 0.1471148615652205, + "instruction_rms": 0.002966617068716632, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.8582763671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 49305.95703125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 400.9898681640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7817898392677307, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31849199533462524, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.1048812866211, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 103.27552032470703, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.6265869140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 758.331298828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1452.0777587890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 400.9898681640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7538063526153564, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3186255693435669, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.0943832397461, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.19519705914213228, + "step": 60, + "teaching_rms": 0.1471148615652205, + "traffic_rms": 0.19517467572955816 + }, + { + "batch_loss": 2.4347341060638428, + "innovation_rms": 0.17018763019291636, + "instruction_rms": 0.003067180835746417, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 173.52566528320312, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 62583.61328125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 484.1830139160156, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7656142115592957, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43213972449302673, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.5843963623047, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 126.43804168701172, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 231.62522888183594, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 931.7066040039062, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1733.7532958984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 484.1831970214844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7376335263252258, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4323510527610779, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.42413330078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.20982407223448857, + "step": 61, + "teaching_rms": 0.17018763019291636, + "traffic_rms": 0.2098020904909738 + }, + { + "batch_loss": 2.5076889991760254, + "innovation_rms": 0.21067832976758744, + "instruction_rms": 0.0032421864834898855, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 237.11221313476562, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 110517.578125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339.7931823730469, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7528095245361328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3302847146987915, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158.31381225585938, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.4746856689453, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280.3663635253906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1139.5257568359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2078.19189453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339.7931823730469, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7248316407203674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33049604296684265, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158.153564453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30778666186090353, + "step": 62, + "teaching_rms": 0.21067832976758744, + "traffic_rms": 0.3077696876069664 + }, + { + "batch_loss": 2.390056848526001, + "innovation_rms": 0.2937789362302948, + "instruction_rms": 0.0033267731235913497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 316.9649963378906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 161927.96875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 745.322265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7433246374130249, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2627495527267456, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 189.59312438964844, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.57167053222656, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 360.96978759765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1411.45751953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2719.31787109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 745.322265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.715349555015564, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2626023590564728, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 189.5918426513672, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3666776037081885, + "step": 63, + "teaching_rms": 0.2937789362302948, + "traffic_rms": 0.36666257920334555 + }, + { + "batch_loss": 2.4369399547576904, + "innovation_rms": 0.30940441221992027, + "instruction_rms": 0.003731943239831233, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 397.1988525390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 225371.90625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 959.9466552734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7385742664337158, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2929586172103882, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.85403442382812, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.13812255859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 425.66448974609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1727.1590576171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3157.015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 959.9468994140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7106019854545593, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29282501339912415, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.69383239746094, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3686243223340822, + "step": 64, + "teaching_rms": 0.30940441221992027, + "traffic_rms": 0.3686053766913465 + }, + { + "batch_loss": 2.524446725845337, + "innovation_rms": 0.48832824542604164, + "instruction_rms": 0.003533626523456973, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.8641052246094, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 405880.53125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 708.9387817382812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7415859699249268, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21936196088790894, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 306.7479248046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 289.77978515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 566.416748046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2158.552490234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4313.93310546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 708.9390258789062, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7136164307594299, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21968209743499756, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 306.5877380371094, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.5558232935480776, + "step": 65, + "teaching_rms": 0.48832824542604164, + "traffic_rms": 0.5558120884985385 + }, + { + "batch_loss": 2.478813648223877, + "innovation_rms": 0.5196817718313895, + "instruction_rms": 0.009888105288152862, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 592.1734619140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 568313.875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 801.5097045898438, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7511019706726074, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30205684900283813, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 337.5060119628906, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 359.656005859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 698.7623291015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2691.443603515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5328.91162109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 801.5097045898438, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.723135232925415, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30198022723197937, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 337.4955139160156, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.6730891562051526, + "step": 66, + "teaching_rms": 0.5196817718313895, + "traffic_rms": 0.6730165866378638 + }, + { + "batch_loss": 2.4153847694396973, + "innovation_rms": 0.6703694776237688, + "instruction_rms": 0.003386271351952712, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 962.3223266601562, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1000257.6875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1159.14453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7644004225730896, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3609061539173126, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 415.41253662109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 446.3525695800781, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 866.9656982421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3353.891845703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6624.48193359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1159.14453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7364364862442017, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3612256944179535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 415.40203857421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8468092911044098, + "step": 67, + "teaching_rms": 0.6703694776237688, + "traffic_rms": 0.84680261079325 + }, + { + "batch_loss": 2.4829368591308594, + "innovation_rms": 0.7444722226012948, + "instruction_rms": 0.002556329548100122, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1155.5069580078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1451303.625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1810.099609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7457849383354187, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3518803119659424, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 517.1397705078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 549.7866821289062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1034.3409423828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4129.1728515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7752.8076171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1810.099609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7178238034248352, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3519507348537445, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 517.1075439453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.0035296590967906, + "step": 68, + "teaching_rms": 0.7444722226012948, + "traffic_rms": 1.0035265351734157 + }, + { + "batch_loss": 2.538494110107422, + "innovation_rms": 0.7005913051484038, + "instruction_rms": 0.0022234507592234594, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1516.2454833984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1836021.25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2883.810546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7133083939552307, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8661357164382935, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 677.12744140625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 661.363525390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1115.768310546875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4949.060546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8198.8779296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2883.810546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6521435976028442, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8665319681167603, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 677.09521484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.0186449001905225, + "step": 69, + "teaching_rms": 0.7005913051484038, + "traffic_rms": 1.018642630245599 + }, + { + "batch_loss": 2.3547451496124268, + "innovation_rms": 0.9249450489321214, + "instruction_rms": 0.0024619186047320535, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2231.51171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5069446.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1884.50341796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7275204658508301, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3429597020149231, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 848.4443969726562, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 789.3265991210938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1279.630859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5866.5849609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9175.2421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1884.50341796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6203043460845947, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3426401615142822, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 848.4121704101562, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.55175802160136, + "step": 70, + "teaching_rms": 0.9249450489321214, + "traffic_rms": 1.5517560431732202 + }, + { + "batch_loss": 2.316836357116699, + "innovation_rms": 1.015477441927235, + "instruction_rms": 0.003318470302719959, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3459.3759765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13310735.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2181.2890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7359165549278259, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3235578238964081, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1027.4617919921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 932.5847778320312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1432.5816650390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6867.78125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10011.9609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2181.2890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6101333498954773, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32330799102783203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1027.4295654296875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.8441746307953975, + "step": 71, + "teaching_rms": 1.015477441927235, + "traffic_rms": 1.844171619938854 + }, + { + "batch_loss": 2.4793853759765625, + "innovation_rms": 1.2470840347143377, + "instruction_rms": 0.0035684630509604815, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5237.3203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 23909672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2817.198486328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7251124978065491, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34493640065193176, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1217.7882080078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1097.4149169921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1648.30078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7998.84130859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11310.6025390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2817.198486328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6172666549682617, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34468531608581543, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1217.7559814453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.1800811128316324, + "step": 72, + "teaching_rms": 1.2470840347143377, + "traffic_rms": 2.1800781892335768 + }, + { + "batch_loss": 2.372307538986206, + "innovation_rms": 1.698276361378599, + "instruction_rms": 0.002923995774828197, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6628.64697265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 31594946.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4588.37890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7110751271247864, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6984946727752686, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1426.06005859375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1301.3475341796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2039.32666015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9405.822265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14069.8056640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4588.37890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6496323347091675, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6980984210968018, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1426.02783203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.5542872903030887, + "step": 73, + "teaching_rms": 1.698276361378599, + "traffic_rms": 2.554285598497763 + }, + { + "batch_loss": 2.3361623287200928, + "innovation_rms": 2.2813399075785536, + "instruction_rms": 0.0027604831903522793, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6564.48779296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34238680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3517.7177734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7091822624206543, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6443103551864624, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1699.24560546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1566.6322021484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2652.84716796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11297.0693359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18912.474609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3517.7177734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6797152161598206, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6439141035079956, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1699.21337890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.6509082692902046, + "step": 74, + "teaching_rms": 2.2813399075785536, + "traffic_rms": 2.650906862568348 + }, + { + "batch_loss": 2.7531442642211914, + "innovation_rms": 2.9659479115635383, + "instruction_rms": 0.009318188013282058, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7956.5341796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 42547996.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3793.674560546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7460945248603821, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4840089678764343, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2014.55908203125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1914.0523681640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3474.20166015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13857.4765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25604.0703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3793.674560546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6418192386627197, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48396971821784973, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2014.52685546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.4773879191707384, + "step": 75, + "teaching_rms": 2.9659479115635383, + "traffic_rms": 3.477375494843261 + }, + { + "batch_loss": 2.3176350593566895, + "innovation_rms": 3.30619803348395, + "instruction_rms": 0.004578899946061204, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.6708984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 45941992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9492.4658203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7816749215126038, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7012515068054199, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2360.502685546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2348.286865234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4342.3447265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17094.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32374.60546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9492.4658203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5878779888153076, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7012122869491577, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2360.470458984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.9499497757210795, + "step": 76, + "teaching_rms": 3.30619803348395, + "traffic_rms": 3.9499471215339197 + }, + { + "batch_loss": 2.401620626449585, + "innovation_rms": 3.416840868412495, + "instruction_rms": 0.004736337879770851, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11468.1572265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 88337336.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5490.1455078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.767420768737793, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4836684465408325, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2856.669189453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2847.915283203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4996.283203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20731.0546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36361.17578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5490.1455078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5522538423538208, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4836292564868927, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2856.636962890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.73412992852912, + "step": 77, + "teaching_rms": 3.416840868412495, + "traffic_rms": 5.734127971765151 + }, + { + "batch_loss": 2.6663780212402344, + "innovation_rms": 3.920423437409811, + "instruction_rms": 0.002911213638411164, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15149.9619140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 159891984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5515.50390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7486695647239685, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40441006422042847, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3408.219482421875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3416.765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5688.50341796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24758.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40270.7109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5515.50390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5508719086647034, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4042114019393921, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3408.187255859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.928286707576013, + "step": 78, + "teaching_rms": 3.920423437409811, + "traffic_rms": 6.928286094742438 + }, + { + "batch_loss": 2.7375292778015137, + "innovation_rms": 5.02494664408494, + "instruction_rms": 0.004562067036786186, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17985.294921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 218299824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9520.3798828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7530592083930969, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3174394965171814, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3966.557861328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4094.35107421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6775.8544921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 29487.5390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47294.140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9520.380859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5753636360168457, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31746843457221985, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3966.525634765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8.181419417227955, + "step": 79, + "teaching_rms": 5.02494664408494, + "traffic_rms": 8.181418154544309 + }, + { + "batch_loss": 2.475745677947998, + "innovation_rms": 7.091648780573026, + "instruction_rms": 0.006264387199601043, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17802.9609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 229880464.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14978.84765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7754802703857422, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5740014314651489, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4754.9462890625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4962.34228515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8679.912109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 35740.73828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62531.98046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14978.84765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6170738935470581, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5742000937461853, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4754.9140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8.330401976660632, + "step": 80, + "teaching_rms": 7.091648780573026, + "traffic_rms": 8.330399624285665 + }, + { + "batch_loss": 2.3406076431274414, + "innovation_rms": 7.726756001041097, + "instruction_rms": 0.007339747084183829, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 22495.953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 291053184.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18461.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7662162184715271, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4867793619632721, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5673.27685546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6003.533203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10411.9091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 43262.41796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75216.7890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18461.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6578763723373413, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4868185520172119, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5673.24462890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10.549765930630656, + "step": 81, + "teaching_rms": 7.726756001041097, + "traffic_rms": 10.549763378890491 + }, + { + "batch_loss": 2.5500175952911377, + "innovation_rms": 10.663997847628384, + "instruction_rms": 0.00780783337556407, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 27122.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 325617248.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24193.84375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7808874845504761, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5599677562713623, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6832.76806640625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7338.189453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13346.564453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 53122.27734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98598.578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24193.84375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6033959984779358, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5597541332244873, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6832.73583984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12.666860925458497, + "step": 82, + "teaching_rms": 10.663997847628384, + "traffic_rms": 12.66685853625757 + }, + { + "batch_loss": 2.7303466796875, + "innovation_rms": 12.933371474751509, + "instruction_rms": 0.006397153697269916, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31366.921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 400405696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20896.29296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7762782573699951, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4770514667034149, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8050.7431640625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9018.892578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16807.02734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 65717.6640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125953.859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20896.29296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5651462078094482, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4768378436565399, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8050.7109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15.122417748197256, + "step": 83, + "teaching_rms": 12.933371474751509, + "traffic_rms": 15.122416397854078 + }, + { + "batch_loss": 2.2869884967803955, + "innovation_rms": 17.35868659804633, + "instruction_rms": 0.006087366650510948, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35008.15625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 458933568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26293.626953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7784696221351624, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3360512852668762, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10555.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11218.359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21994.66796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 82541.484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168238.203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26293.626953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5686003565788269, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.336360365152359, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10555.0751953125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19.22030932351916, + "step": 84, + "teaching_rms": 17.35868659804633, + "traffic_rms": 19.220308355817977 + }, + { + "batch_loss": 2.417844772338867, + "innovation_rms": 18.111740999657368, + "instruction_rms": 0.005105615205234677, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43919.55078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 665333440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22361.314453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7812217473983765, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5540875792503357, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12195.359375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13803.4169921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25850.578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 102265.046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197235.609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22361.314453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5669358968734741, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5543966293334961, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12195.3720703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23.474712817266216, + "step": 85, + "teaching_rms": 18.111740999657368, + "traffic_rms": 23.474712259349154 + }, + { + "batch_loss": 2.34238338470459, + "innovation_rms": 26.643317652450072, + "instruction_rms": 0.00789851258972025, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 46877.578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1045613056.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57101.90625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7764147520065308, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4425278604030609, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15933.095703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17159.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33557.70703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 128206.5078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 259414.609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57101.90625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.562512218952179, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4428368806838989, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15933.0634765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 29.975651468161903, + "step": 86, + "teaching_rms": 26.643317652450072, + "traffic_rms": 29.97565042630799 + }, + { + "batch_loss": 2.5814154148101807, + "innovation_rms": 25.31231570706793, + "instruction_rms": 0.007001258486654383, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 71465.171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2045940864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63003.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.788983166217804, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49924302101135254, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22233.478515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20977.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38182.4921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 157079.28125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288727.75, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63003.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5670055150985718, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4989340305328369, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22233.447265625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 39.47756710021245, + "step": 87, + "teaching_rms": 25.31231570706793, + "traffic_rms": 39.47756648275737 + }, + { + "batch_loss": 2.3756399154663086, + "innovation_rms": 27.542507437195862, + "instruction_rms": 0.007025563112211757, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 103843.2265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5197708288.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71646.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8043374419212341, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6340539455413818, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 29398.166015625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25208.076171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42306.38671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 188182.21875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311029.4375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71646.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5740029811859131, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.633745014667511, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 29398.134765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 47.69299100093232, + "step": 88, + "teaching_rms": 27.542507437195862, + "traffic_rms": 47.69299048587187 + }, + { + "batch_loss": 2.3292248249053955, + "innovation_rms": 34.04057521454342, + "instruction_rms": 0.00928481384583604, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 131851.453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8641670144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76690.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8141071796417236, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4324672222137451, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 37067.2265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30041.89453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48338.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 222840.828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 346586.03125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76690.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5827139616012573, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43215829133987427, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 37067.1953125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 57.34669496474009, + "step": 89, + "teaching_rms": 34.04057521454342, + "traffic_rms": 57.34669421449559 + }, + { + "batch_loss": 2.4141571521759033, + "innovation_rms": 48.443801918475934, + "instruction_rms": 0.007666663413763363, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 145390.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10646313984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97706.4296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8251636028289795, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22993513941764832, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44619.12109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 36079.8359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60379.421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 266116.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 432758.59375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97706.4296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.589754045009613, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2299315631389618, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44619.08984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 67.83430556002399, + "step": 90, + "teaching_rms": 48.443801918475934, + "traffic_rms": 67.83430512686907 + }, + { + "batch_loss": 2.405642032623291, + "innovation_rms": 58.37057487448977, + "instruction_rms": 0.005456265539871574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 130791.21875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11602495488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170226.78125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8396300673484802, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24015793204307556, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61641.80078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 43532.11328125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74522.765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 319902.40625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 537857.125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170226.78125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5937805771827698, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24014492332935333, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61641.76953125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 70.72122174316766, + "step": 91, + "teaching_rms": 58.37057487448977, + "traffic_rms": 70.7212215322067 + }, + { + "batch_loss": 2.3483259677886963, + "innovation_rms": 71.97660133175926, + "instruction_rms": 0.004971039328885624, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 187681.921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19997165568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190243.84375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8390191197395325, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38936614990234375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80666.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 52689.8828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91577.6796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 386464.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 665622.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190243.84375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5970160961151123, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3893531560897827, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80666.15625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 94.74296018398118, + "step": 92, + "teaching_rms": 71.97660133175926, + "traffic_rms": 94.74296005374006 + }, + { + "batch_loss": 2.3587899208068848, + "innovation_rms": 91.49431086729045, + "instruction_rms": 0.004603780988929752, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 264642.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 25319251968.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195796.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8269176483154297, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3728739619255066, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 100245.796875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 64308.76953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116188.8671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 472283.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858190.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195796.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5925329923629761, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3730345666408539, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 100245.765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 111.9506968983813, + "step": 93, + "teaching_rms": 91.49431086729045, + "traffic_rms": 111.95069680362793 + }, + { + "batch_loss": 2.3371973037719727, + "innovation_rms": 118.88152638512688, + "instruction_rms": 0.007733715261588085, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 323254.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34859601920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176584.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8154491186141968, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3346499800682068, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117904.2578125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 79284.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 149759.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 585442.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1131590.625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176584.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.591230571269989, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3346535563468933, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117904.2265625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 139.0453557111831, + "step": 94, + "teaching_rms": 118.88152638512688, + "traffic_rms": 139.04535549672562 + }, + { + "batch_loss": 2.3224785327911377, + "innovation_rms": 130.24339092732598, + "instruction_rms": 0.0041314551897527856, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 365733.40625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 56867098624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213951.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8138909339904785, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4847105145454407, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 135029.328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 97314.5078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 180298.171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 720154.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1347117.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213951.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5885355472564697, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48438751697540283, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 135029.296875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 180.41483420942686, + "step": 95, + "teaching_rms": 130.24339092732598, + "traffic_rms": 180.414834162654 + }, + { + "batch_loss": 2.55312180519104, + "innovation_rms": 162.15771491626964, + "instruction_rms": 0.003997404185917541, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 330812.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66318614528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 383270.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8451739549636841, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.385018527507782, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 173356.390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 119320.609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220061.03125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 884211.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1640573.375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 383270.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.588748574256897, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38469555974006653, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 173356.359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 189.89317515110562, + "step": 96, + "teaching_rms": 162.15771491626964, + "traffic_rms": 189.8931751096196 + }, + { + "batch_loss": 2.406096935272217, + "innovation_rms": 191.94311349046978, + "instruction_rms": 0.0043017700827133595, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 456613.71875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 119878754304.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 415717.53125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8708974719047546, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4519978165626526, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214928.140625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 145625.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 263047.90625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1079009.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1947971.875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 415717.53125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5959071516990662, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4520557224750519, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214928.109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 264.4544270791255, + "step": 97, + "teaching_rms": 191.94311349046978, + "traffic_rms": 264.45442704411784 + }, + { + "batch_loss": 2.3980355262756348, + "innovation_rms": 270.7133040073918, + "instruction_rms": 0.005613491132458989, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 561907.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 167811203072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 527557.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8600472211837769, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30987223982810974, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 267683.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 179302.578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 336771.65625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1332344.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2533354.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 527557.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6075604557991028, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30998218059539795, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 267683.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 325.7387253463419, + "step": 98, + "teaching_rms": 270.7133040073918, + "traffic_rms": 325.73872529794295 + }, + { + "batch_loss": 2.511371374130249, + "innovation_rms": 308.6624095161421, + "instruction_rms": 0.00563347746209227, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 552709.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 202030153728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 475708.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8443896770477295, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5333326458930969, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 309883.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 221055.859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 417532.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1644415.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3120706.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 475708.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6098958849906921, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5332561731338501, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 309883.28125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 374.97789534222545, + "step": 99, + "teaching_rms": 308.6624095161421, + "traffic_rms": 374.97789529982776 + }, + { + "batch_loss": 2.499373197555542, + "innovation_rms": 276.4300086022807, + "instruction_rms": 0.00795094610111459, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 610148.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 282602209280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1033069.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8406844735145569, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36782869696617126, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 411329.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 266660.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 456046.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1975061.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3306460.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1033069.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6007090210914612, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36814743280410767, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 411329.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 432.79271494139215, + "step": 100, + "teaching_rms": 276.4300086022807, + "traffic_rms": 432.79271486842003 + }, + { + "batch_loss": 2.529513120651245, + "innovation_rms": 370.1983623302733, + "instruction_rms": 0.005114144355957582, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1046351.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 716328009728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 931162.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8500613570213318, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40649712085723877, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 504446.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 319058.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 523979.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2347177.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3721158.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 931162.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5929791927337646, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40617430210113525, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 504446.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 611.4207484338765, + "step": 101, + "teaching_rms": 370.1983623302733, + "traffic_rms": 611.4207484126694 + }, + { + "batch_loss": 2.7679178714752197, + "innovation_rms": 485.69120123561584, + "instruction_rms": 0.004764945281057927, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1540122.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1389466091520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 917895.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8667166829109192, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7315129637718201, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 596235.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 382022.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 629640.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2789865.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4426884.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 917895.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5979268550872803, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7316721677780151, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 596235.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 713.7298082700961, + "step": 102, + "teaching_rms": 485.69120123561584, + "traffic_rms": 713.7298082545384 + }, + { + "batch_loss": 2.655229330062866, + "innovation_rms": 654.1948078071026, + "instruction_rms": 0.009705297947522068, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1816187.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1561586434048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1163023.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8849784731864929, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4897858202457428, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 605394.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 462408.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 803859.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3369735.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5798702.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1163023.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6122010946273804, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4899449944496155, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 605394.3125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 798.3300693407525, + "step": 103, + "teaching_rms": 654.1948078071026, + "traffic_rms": 798.3300692818362 + }, + { + "batch_loss": 2.777411699295044, + "innovation_rms": 775.5768295324601, + "instruction_rms": 0.005402065788018852, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2620166.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2577360748544.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 918536.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8656856417655945, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5031324625015259, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 653770.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 560651.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 982425.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4084135.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7143995.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 918536.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6330015659332275, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5033369064331055, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 653770.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1051.3147617876177, + "step": 104, + "teaching_rms": 775.5768295324601, + "traffic_rms": 1051.3147617737145 + }, + { + "batch_loss": 2.7247321605682373, + "innovation_rms": 1003.2032933793764, + "instruction_rms": 0.004578440599285676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3314414.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3536481157120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1537840.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8517073392868042, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.359123170375824, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 734321.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 684527.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1238768.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4998846.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9147112.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1537840.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6209098100662231, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3593275845050812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 734321.4375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1260.3692178474582, + "step": 105, + "teaching_rms": 1003.2032933793764, + "traffic_rms": 1260.3692178391175 + }, + { + "batch_loss": 3.032888174057007, + "innovation_rms": 1004.467850751781, + "instruction_rms": 0.012257602593366775, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3174741.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4047758950400.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1943421.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.849386990070343, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6818010807037354, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 709045.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 828569.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1440413.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6050832.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10519858.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1943421.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6036407351493835, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6815967559814453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 709045.1875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1304.9759910294424, + "step": 106, + "teaching_rms": 1004.467850751781, + "traffic_rms": 1304.9759909719903 + }, + { + "batch_loss": 2.604095697402954, + "innovation_rms": 1267.3414230721685, + "instruction_rms": 0.005072091846888826, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4493724.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7574885761024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1444762.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8594287633895874, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7318201065063477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 704035.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 998740.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1701708.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281228.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12303954.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1444762.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5957499742507935, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7318071126937866, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 704035.625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1867.0515118513474, + "step": 107, + "teaching_rms": 1267.3414230721685, + "traffic_rms": 1867.0515118443434 + }, + { + "batch_loss": 2.6522295475006104, + "innovation_rms": 1508.232553284179, + "instruction_rms": 0.0018199010236073228, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6027123.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12484941447168.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2163377.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8770179152488708, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4229031503200531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 856899.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1199385.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2006452.625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720405.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14391769.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2163377.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6031433343887329, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42294013500213623, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 856899.625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2259.303798128027, + "step": 108, + "teaching_rms": 1508.232553284179, + "traffic_rms": 2259.3037981273606 + }, + { + "batch_loss": 2.952605962753296, + "innovation_rms": 2373.2118485852848, + "instruction_rms": 0.002126473447764236, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6607756.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14608518610944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3258758.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8942528963088989, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5423853993415833, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1056765.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1468899.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2695141.25, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10726995.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20065900.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3258758.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6226229667663574, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5424191951751709, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1056765.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2672.21375415414, + "step": 109, + "teaching_rms": 2373.2118485852848, + "traffic_rms": 2672.213754153368 + }, + { + "batch_loss": 2.987191915512085, + "innovation_rms": 1711.4164648745655, + "instruction_rms": 0.0026689316617151283, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6385705.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 16571441348608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3902357.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8709781169891357, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5321722030639648, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1348390.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1761028.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2921291.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12835236.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21082410.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3902357.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.639066755771637, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5322364568710327, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1348390.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2940.085310071852, + "step": 110, + "teaching_rms": 1711.4164648745655, + "traffic_rms": 2940.085310070686 + }, + { + "batch_loss": 3.106152057647705, + "innovation_rms": 2749.891580899507, + "instruction_rms": 0.003192895175902622, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6843625.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 18298026590208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6457168.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8580403327941895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8518166542053223, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852875.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2121264.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3602359.25, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15437154.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26019180.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6457168.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6144711375236511, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8515827059745789, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852875.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3734.24202192734, + "step": 111, + "teaching_rms": 2749.891580899507, + "traffic_rms": 3734.242021925974 + }, + { + "batch_loss": 2.5321998596191406, + "innovation_rms": 1677.0369304751378, + "instruction_rms": 0.002966719285873498, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7692579.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24491260379136.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7002883.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8519228100776672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4954575002193451, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2553163.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2479267.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3580025.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17947436.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25102814.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7002883.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.624965488910675, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49522361159324646, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2553163.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3572.6393726514507, + "step": 112, + "teaching_rms": 1677.0369304751378, + "traffic_rms": 3572.6393726502447 + }, + { + "batch_loss": 3.113913059234619, + "innovation_rms": 3450.639887395024, + "instruction_rms": 0.002922545772050476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10595736.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34484126744576.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10025976.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8609117865562439, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6340064406394958, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3555761.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2919057.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4397905.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20969312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30218756.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10025976.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6078451871871948, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6339882612228394, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3555761.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4924.615597452094, + "step": 113, + "teaching_rms": 3450.639887395024, + "traffic_rms": 4924.615597451271 + }, + { + "batch_loss": 2.8277428150177, + "innovation_rms": 4024.8570265599983, + "instruction_rms": 0.003173735611237706, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14566514.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80572890218496.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8564988.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8844509720802307, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5984358787536621, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4412260.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3451283.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5322258.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24577612.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36082992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8564988.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6099280118942261, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5984377861022949, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4412260.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6290.455261057078, + "step": 114, + "teaching_rms": 4024.8570265599983, + "traffic_rms": 6290.455261056299 + }, + { + "batch_loss": 2.766347885131836, + "innovation_rms": 4348.6102105058335, + "instruction_rms": 0.002795482458835721, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19252634.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 174793659777024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7477092.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9196439385414124, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7779839634895325, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5153251.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4073074.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6217907.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 28696384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41187720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7477092.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6451484560966492, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7780022025108337, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5153251.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7102.298299176513, + "step": 115, + "teaching_rms": 4348.6102105058335, + "traffic_rms": 7102.298299175982 + }, + { + "batch_loss": 2.875523328781128, + "innovation_rms": 5546.76149611583, + "instruction_rms": 0.004893143660741044, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24882764.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 282616666259456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7210041.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9218015074729919, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6897612810134888, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5800901.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4823745.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7506714.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 33647276.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49508904.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7210041.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6473335027694702, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.689667820930481, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5800901.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8326.33455552324, + "step": 116, + "teaching_rms": 5546.76149611583, + "traffic_rms": 8326.334555521826 + }, + { + "batch_loss": 3.3464300632476807, + "innovation_rms": 7811.099022525484, + "instruction_rms": 0.012659903845633232, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31524012.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 358589805887488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12823224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8920026421546936, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5220457315444946, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6342772.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5790217.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9664716.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 40285884.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66386080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12823224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6175620555877686, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5223590135574341, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6342772.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9937.345599576454, + "step": 117, + "teaching_rms": 7811.099022525484, + "traffic_rms": 9937.34559956838 + }, + { + "batch_loss": 2.7977757453918457, + "innovation_rms": 8879.631586202338, + "instruction_rms": 0.004945157847004095, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37892340.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 403184921083904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15194471.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8796817064285278, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6608639359474182, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7454141.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6980519.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11903023.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 48623140.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83372544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15194471.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6052685976028442, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6609573364257812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7454141.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10560.231156893973, + "step": 118, + "teaching_rms": 8879.631586202338, + "traffic_rms": 10560.231156892834 + }, + { + "batch_loss": 3.032102108001709, + "innovation_rms": 11758.656987866265, + "instruction_rms": 0.002802772933752446, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 49819200.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 539399641104384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16340328.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8839819431304932, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5893049240112305, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9088174.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8497031.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15165113.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 59562488.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109393464.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16340328.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.609596312046051, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5894078016281128, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9088174.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14494.79512880959, + "step": 119, + "teaching_rms": 11758.656987866265, + "traffic_rms": 14494.795128809332 + }, + { + "batch_loss": 2.678724765777588, + "innovation_rms": 14800.13548458601, + "instruction_rms": 0.0030559744191655316, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 58607752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 644219660140544.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17658252.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9034600257873535, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9160849452018738, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9833844.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10435391.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19383596.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 73890816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 143283248.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17658252.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6291018128395081, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9161878824234009, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9833844.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17027.364763306658, + "step": 120, + "teaching_rms": 14800.13548458601, + "traffic_rms": 17027.364763306392 + }, + { + "batch_loss": 2.5990915298461914, + "innovation_rms": 14055.496697281107, + "instruction_rms": 0.0032957353605324994, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 71424048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 793933227491328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33801208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8512734770774841, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1153209209442139, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12267997.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12651133.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22157424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 89914720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160239072.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33801208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5769427418708801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1154237985610962, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12267997.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 18993.85507725448, + "step": 121, + "teaching_rms": 14055.496697281107, + "traffic_rms": 18993.85507725419 + }, + { + "batch_loss": 3.113199234008789, + "innovation_rms": 19656.663178643117, + "instruction_rms": 0.0031283461927967588, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 90680536.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1250661727469568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29707268.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8328803181648254, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9395201802253723, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15041356.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15348700.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26975674.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 109439480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195247584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29707268.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5938446521759033, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9392022490501404, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15041356.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26825.51216902535, + "step": 122, + "teaching_rms": 19656.663178643117, + "traffic_rms": 26825.512169025187 + }, + { + "batch_loss": 2.956714391708374, + "innovation_rms": 25768.536693021757, + "instruction_rms": 0.003992936719930422, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104680720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1628125465149440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53450040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8298673629760742, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0616412162780762, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16814664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18745848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33971480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 134353392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249139072.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53450040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6277175545692444, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0619478225708008, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16814664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31169.98111617491, + "step": 123, + "teaching_rms": 25768.536693021757, + "traffic_rms": 31169.98111617466 + }, + { + "batch_loss": 2.942202568054199, + "innovation_rms": 22111.31353618656, + "instruction_rms": 0.005518233511845173, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 120016816.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2145930212016128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53497908.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8274825215339661, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7296820282936096, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20398286.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 22495226.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37493772.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 160934976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 265815824.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53497908.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5841032266616821, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7299886345863342, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20398286.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31765.848084410354, + "step": 124, + "teaching_rms": 22111.31353618656, + "traffic_rms": 31765.848084409878 + }, + { + "batch_loss": 2.8347885608673096, + "innovation_rms": 38003.609428714044, + "instruction_rms": 0.00432232167716717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 151779760.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3070609827299328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65027476.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8186516165733337, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.535667359828949, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25012360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 27257160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47619348.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 195701664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 347666816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65027476.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5880117416381836, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5358380675315857, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25012360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 47319.435438949906, + "step": 125, + "teaching_rms": 38003.609428714044, + "traffic_rms": 47319.4354389497 + }, + { + "batch_loss": 2.9453041553497314, + "innovation_rms": 42738.49152014467, + "instruction_rms": 0.004335329185343499, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 163649824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4202251836981248.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85264376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8030478358268738, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3768886923789978, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33538798.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33081402.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58242416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 238238144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 425364864.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85264376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5971006751060486, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37686002254486084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33538798.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52884.55727912229, + "step": 126, + "teaching_rms": 42738.49152014467, + "traffic_rms": 52884.557279122106 + }, + { + "batch_loss": 2.997237205505371, + "innovation_rms": 61421.85592154468, + "instruction_rms": 0.004081904971423547, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 192242112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5975407610298368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62751712.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7920379638671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5090305805206299, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 39601424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 40759744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76783424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 295909216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 576710784.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62751712.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6040566563606262, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5092204213142395, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 39601424.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 69285.43577241883, + "step": 127, + "teaching_rms": 61421.85592154468, + "traffic_rms": 69285.43577241871 + }, + { + "batch_loss": 2.858107089996338, + "innovation_rms": 73564.8650253413, + "instruction_rms": 0.0035219038066768836, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 225553568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9251229632823296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137278000.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.78589928150177, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6522449851036072, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 50197848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 50536308.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97765640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 370144256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 742350272.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137278000.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6092165112495422, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6523159146308899, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 50197848.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 85161.19723871013, + "step": 128, + "teaching_rms": 73564.8650253413, + "traffic_rms": 85161.19723871005 + }, + { + "batch_loss": 2.57977294921875, + "innovation_rms": 64679.86530338011, + "instruction_rms": 0.0032553503110385115, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 260530464.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4300036703715328e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94830752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7881525158882141, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7665858864784241, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58563488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 61479392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109430848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 449729184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 795849216.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94830752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6117116808891296, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7666568160057068, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58563488.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 118022.60111594712, + "step": 129, + "teaching_rms": 64679.86530338011, + "traffic_rms": 118022.60111594708 + }, + { + "batch_loss": 2.3446245193481445, + "innovation_rms": 86226.83434427669, + "instruction_rms": 0.0028335145735623327, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 322197312.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3387101949067264e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168791008.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7926082611083984, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6233149170875549, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67184552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 74314984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128355952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 540377920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 906487360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168791008.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6135299205780029, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6231443285942078, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67184552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 146895.25298749757, + "step": 130, + "teaching_rms": 86226.83434427669, + "traffic_rms": 146895.25298749754 + }, + { + "batch_loss": 2.816110134124756, + "innovation_rms": 122196.68763312898, + "instruction_rms": 0.002662240826493286, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 411817568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.092627923455181e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 274677120.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7979832291603088, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35858768224716187, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83716992.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 90392288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160773056.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 654978816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1146008960.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 274677120.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6151235103607178, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.358634352684021, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83716992.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 158170.68211518423, + "step": 131, + "teaching_rms": 122196.68763312898, + "traffic_rms": 158170.6821151842 + }, + { + "batch_loss": 2.4687514305114746, + "innovation_rms": 103057.31745808628, + "instruction_rms": 0.004291214368352755, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 439368288.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.003515969254195e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 246004576.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8130159974098206, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5470203161239624, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 94986280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 107768912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 173766224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 776864640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1218858368.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 246004576.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6127592325210571, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5471043586730957, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 94986280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 145550.15609434096, + "step": 132, + "teaching_rms": 103057.31745808628, + "traffic_rms": 145550.1560943409 + }, + { + "batch_loss": 2.512118339538574, + "innovation_rms": 168947.2542805254, + "instruction_rms": 0.003241834585915969, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 613254720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.68581219343401e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 284436960.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8202466368675232, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5174924731254578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106206976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 129438152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 216692384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 930117504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1532528384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 284436960.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6129554510116577, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5176523923873901, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106206976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 239006.05668549045, + "step": 133, + "teaching_rms": 168947.2542805254, + "traffic_rms": 239006.05668549045 + }, + { + "batch_loss": 2.4463486671447754, + "innovation_rms": 205828.8760554794, + "instruction_rms": 0.0031027911945544543, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 807217920.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.272038316572672e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 451258048.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8161686062812805, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47893109917640686, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 129742944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 156320688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 268825408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1122261504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1921439744.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 451258048.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6184041500091553, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47909098863601685, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 129742944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 277244.1890132393, + "step": 134, + "teaching_rms": 205828.8760554794, + "traffic_rms": 277244.1890132392 + }, + { + "batch_loss": 2.4676835536956787, + "innovation_rms": 259553.55076245498, + "instruction_rms": 0.004408668335742819, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 906798720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5323127334148506e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 537525056.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8213427662849426, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43375200033187866, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158767728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 190287872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339671904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1368372608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2461110784.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 537525056.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.618796706199646, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4336836338043213, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158767728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 298099.7396158965, + "step": 135, + "teaching_rms": 259553.55076245498, + "traffic_rms": 298099.7396158964 + }, + { + "batch_loss": 2.675889015197754, + "innovation_rms": 270831.96838941635, + "instruction_rms": 0.0037714932158810087, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1338274432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4114447631764685e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 426337408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8393154740333557, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3410657048225403, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 198973664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 229712256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 394243808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1652490752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2841181184.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 426337408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6105753779411316, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34124788641929626, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 198973664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 403808.4939087168, + "step": 136, + "teaching_rms": 270831.96838941635, + "traffic_rms": 403808.4939087168 + }, + { + "batch_loss": 2.416773796081543, + "innovation_rms": 315599.1833513622, + "instruction_rms": 0.004232996476216561, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1841195904.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.434012192256819e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 526804928.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8563364744186401, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6072121858596802, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 251654160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 275380768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 456684960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1977917568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3254268160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 526804928.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6036926507949829, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6072804927825928, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 251654160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 490629.2464427476, + "step": 137, + "teaching_rms": 315599.1833513622, + "traffic_rms": 490629.24644274754 + }, + { + "batch_loss": 2.388036012649536, + "innovation_rms": 439833.5132235684, + "instruction_rms": 0.004660248757535669, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2204904704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.211005776550953e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 984912896.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8497956991195679, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43208634853363037, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 317994464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 332349248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 569684800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2387376896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4094593536.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 984912896.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6177691221237183, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4321546256542206, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 317994464.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 590013.331614763, + "step": 138, + "teaching_rms": 439833.5132235684, + "traffic_rms": 590013.3316147629 + }, + { + "batch_loss": 2.638763904571533, + "innovation_rms": 395493.1607665062, + "instruction_rms": 0.0078084766640674095, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2130334592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.07739001547391e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1040332160.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8472342491149902, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2820151150226593, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 379701920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 395101984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 627527360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2832377600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4450007040.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1040332160.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6278207302093506, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2820531129837036, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 379701920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 553212.286902753, + "step": 139, + "teaching_rms": 395493.1607665062, + "traffic_rms": 553212.286902753 + }, + { + "batch_loss": 2.522965431213379, + "innovation_rms": 598313.9322366331, + "instruction_rms": 0.005855539019422693, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2725411072.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2545710797168312e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 807673984.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8471391201019287, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35713261365890503, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 458239968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 472170912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 770689216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3382531072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5501534208.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 807673984.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6343858242034912, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3571133613586426, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 458239968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 825260.6455647685, + "step": 140, + "teaching_rms": 598313.9322366331, + "traffic_rms": 825260.6455647685 + }, + { + "batch_loss": 2.2690601348876953, + "innovation_rms": 765368.829244535, + "instruction_rms": 0.007787019182553191, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3244834048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5915749670429655e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1915923584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.847508430480957, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3862747848033905, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 574574912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 569239360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 970684416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4085069056.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7025380352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1915923584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6397769451141357, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38625550270080566, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 574574912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1002325.6416308341, + "step": 141, + "teaching_rms": 765368.829244535, + "traffic_rms": 1002325.641630834 + }, + { + "batch_loss": 2.326677083969116, + "innovation_rms": 521551.3145758285, + "instruction_rms": 0.006820545432582922, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2970841088.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9376551854174372e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111312512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8617931008338928, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5231277942657471, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 681511616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 668550080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 993106944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4789255680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7041866752.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111312512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6181994080543518, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5229958295822144, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 681511616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 908176.2871045407, + "step": 142, + "teaching_rms": 521551.3145758285, + "traffic_rms": 908176.2871045407 + }, + { + "batch_loss": 2.4174654483795166, + "innovation_rms": 1137200.5953324845, + "instruction_rms": 0.006013736276698576, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3360137472.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2894153434220462e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1990149632.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.871753990650177, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.473938524723053, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 702365376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 800696448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1321463552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5751250432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9619949568.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1990149632.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6043559312820435, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47390052676200867, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 702365376.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1328888.7044480224, + "step": 143, + "teaching_rms": 1137200.5953324845, + "traffic_rms": 1328888.7044480224 + }, + { + "batch_loss": 2.403341293334961, + "innovation_rms": 1184605.6599387259, + "instruction_rms": 0.006567230912498055, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4262600960.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3495461755428536e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1946561536.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8695893883705139, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5519634485244751, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 892768576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 959907520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1592110848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6904283648.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11530332160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1946561536.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6147809624671936, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.552095353603363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 892768576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1691388.3605626293, + "step": 144, + "teaching_rms": 1184605.6599387259, + "traffic_rms": 1691388.3605626293 + }, + { + "batch_loss": 2.3383758068084717, + "innovation_rms": 1574716.2428951699, + "instruction_rms": 0.009563368259102109, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4700674048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.638028667761459e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2964750848.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8629785776138306, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4211433231830597, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1175982336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1160368640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2004610816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8376384512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14721007616.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2964750848.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6308902502059937, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42116260528564453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1175982336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1906418.460484727, + "step": 145, + "teaching_rms": 1574716.2428951699, + "traffic_rms": 1906418.460484727 + }, + { + "batch_loss": 2.553907871246338, + "innovation_rms": 1710705.1630244497, + "instruction_rms": 0.005969184651601473, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5739469824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.833596158267359e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5133174784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.858403742313385, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45573514699935913, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1507362432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1398475008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2381063680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10114867200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17384828928.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5133174784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6429369449615479, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45573702454566956, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1507362432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2296843.791525927, + "step": 146, + "teaching_rms": 1710705.1630244497, + "traffic_rms": 2296843.791525927 + }, + { + "batch_loss": 2.3139073848724365, + "innovation_rms": 1740481.8239120243, + "instruction_rms": 0.005496921432506019, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7201754112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1672777179795554e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3114700288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8584414720535278, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3880077302455902, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1761410816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1668514304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2700393472.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12020429824.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19055628288.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3114700288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6429403424263, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3880457282066345, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1761410816.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2721800.629951973, + "step": 147, + "teaching_rms": 1740481.8239120243, + "traffic_rms": 2721800.629951973 + }, + { + "batch_loss": 2.2919819355010986, + "innovation_rms": 2162801.2408423643, + "instruction_rms": 0.0038778856078143624, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9009761280.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0757374356550582e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4506226688.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8587350249290466, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3419153094291687, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1963210880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1982428928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3139145728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14182696960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21622673408.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4506226688.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6408703923225403, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34178346395492554, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1963210880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3273507.423747624, + "step": 148, + "teaching_rms": 2162801.2408423643, + "traffic_rms": 3273507.423747624 + }, + { + "batch_loss": 2.1796751022338867, + "innovation_rms": 3326609.233870736, + "instruction_rms": 0.00469580924084197, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11057815552.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.602900484615517e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9099461632.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8629697561264038, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2408234179019928, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2291133696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2389276416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4068474624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17067377664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28846804992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9099461632.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6348426938056946, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24069160223007202, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2291133696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4313688.775824422, + "step": 149, + "teaching_rms": 3326609.233870736, + "traffic_rms": 4313688.775824422 + }, + { + "batch_loss": 2.5520925521850586, + "innovation_rms": 1820715.1190012635, + "instruction_rms": 0.010972675389110354, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11424176128.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5837695300986733e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5185865216.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8703330159187317, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4153241515159607, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2809720320.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2786427648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3971512320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19857534976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27901562880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5185865216.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6203871369361877, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41535913944244385, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2809720320.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3573756.9686367647, + "step": 150, + "teaching_rms": 1820715.1190012635, + "traffic_rms": 3573756.9686367647 + }, + { + "batch_loss": 2.445791244506836, + "innovation_rms": 3239531.508001368, + "instruction_rms": 0.0035574149490026296, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13655975936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.6340329325549584e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10046334976.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8529826998710632, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4077705442905426, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3578320896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3240926976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4544994304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22962446336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31049117696.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10046334976.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6244294047355652, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.407733678817749, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3578320896.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5075644.877113855, + "step": 151, + "teaching_rms": 3239531.508001368, + "traffic_rms": 5075644.877113855 + }, + { + "batch_loss": 2.46712064743042, + "innovation_rms": 4063988.440134606, + "instruction_rms": 0.0022735639119511584, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20616591360.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2621691558645308e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9407627264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8395338654518127, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3546615242958069, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4366346240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3777802496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5368754176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 26561509376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35990638592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9407627264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6275916695594788, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3545854687690735, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4366346240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7081068.121418759, + "step": 152, + "teaching_rms": 4063988.440134606, + "traffic_rms": 7081068.121418759 + }, + { + "batch_loss": 2.4161312580108643, + "innovation_rms": 4092126.0233482863, + "instruction_rms": 0.0018124843172816266, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29818744832.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.419172947283862e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9165213696.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8308744430541992, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6174152493476868, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5172312576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4376985600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5991828480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 30468014080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39065038848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9165213696.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6301289796829224, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.617428183555603, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5172312576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8370095.7530873865, + "step": 153, + "teaching_rms": 4092126.0233482863, + "traffic_rms": 8370095.7530873865 + }, + { + "batch_loss": 2.3446731567382812, + "innovation_rms": 4591483.266929219, + "instruction_rms": 0.0016859995113782484, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 39954276352.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.502329511619844e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8949954560.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331393599510193, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4989217519760132, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6002057216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5044373504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6673878016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 34707881984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42398670848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8949954560.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6320238709449768, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49893465638160706, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6002057216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9655416.807834249, + "step": 154, + "teaching_rms": 4591483.266929219, + "traffic_rms": 9655416.807834249 + }, + { + "batch_loss": 2.7820374965667725, + "innovation_rms": 5383696.412478801, + "instruction_rms": 0.0021308357348830574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 50050330624.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.751777270071831e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8857430016.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8457226753234863, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3360220789909363, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6883958272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5798594048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7542202880.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 39409496064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47016136704.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8857430016.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6330896615982056, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3362038731575012, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6883958272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10901600.095127119, + "step": 155, + "teaching_rms": 5383696.412478801, + "traffic_rms": 10901600.095127119 + }, + { + "batch_loss": 2.8582613468170166, + "innovation_rms": 6550366.478823097, + "instruction_rms": 0.002893179740879398, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 59280908288.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2325783600660092e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10319353856.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8691279292106628, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5364248156547546, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7853398016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6675681792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770876416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 44828258304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54187601920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10319353856.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6333646178245544, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.536465048789978, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7853398016.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12408565.594883475, + "step": 156, + "teaching_rms": 6550366.478823097, + "traffic_rms": 12408565.594883475 + }, + { + "batch_loss": 2.6353585720062256, + "innovation_rms": 8568141.417016175, + "instruction_rms": 0.002959583562323324, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 67050127360.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3625454899124807e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17127469056.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8889334201812744, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46317780017852783, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8945551360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7750356992.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10746752000.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 51575091200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67468328960.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17127469056.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6337318420410156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46298736333847046, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8945551360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14065278.261068987, + "step": 157, + "teaching_rms": 8568141.417016175, + "traffic_rms": 14065278.261068987 + }, + { + "batch_loss": 2.807147979736328, + "innovation_rms": 12616638.395217277, + "instruction_rms": 0.0042111796349499474, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 71937638400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3393732031922643e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22398599168.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8477557897567749, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8760837316513062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10107452416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9199086592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14487293952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 61397295104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98222047232.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22398599168.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6377211213111877, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8759511709213257, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10107452416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14820755.875709733, + "step": 158, + "teaching_rms": 12616638.395217277, + "traffic_rms": 14820755.875709733 + }, + { + "batch_loss": 2.4619638919830322, + "innovation_rms": 7768209.090453337, + "instruction_rms": 0.0030700269283553605, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 71615553536.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4038665794435584e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14932363264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8214743733406067, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5674369931221008, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11173695488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10677000192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14779133952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 71304167424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99068706816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14932363264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6501266956329346, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5673044323921204, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11173695488.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13418884.78542945, + "step": 159, + "teaching_rms": 7768209.090453337, + "traffic_rms": 13418884.78542945 + }, + { + "batch_loss": 2.99658465385437, + "innovation_rms": 13318571.813975804, + "instruction_rms": 0.0038086512546863216, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 75782496256.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4564361127314357e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37489790976.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8085182905197144, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.443156898021698, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12022502400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12453099520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17760989184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 83144523776.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118403555328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37489790976.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6627016067504883, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44318661093711853, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12022502400.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 18153383.278140858, + "step": 160, + "teaching_rms": 13318571.813975804, + "traffic_rms": 18153383.278140858 + }, + { + "batch_loss": 2.7394261360168457, + "innovation_rms": 15455111.600868184, + "instruction_rms": 0.005129046023669375, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 92241969152.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2545222396599935e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29765263360.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8099220991134644, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.444038987159729, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14999028736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 14555921408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21028214784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 97050083328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139055595520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29765263360.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6708148121833801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44406870007514954, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14999028736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26499166.366656564, + "step": 161, + "teaching_rms": 15455111.600868184, + "traffic_rms": 26499166.366656564 + }, + { + "batch_loss": 2.59572172164917, + "innovation_rms": 15457191.931064062, + "instruction_rms": 0.005151584331422955, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 115073703936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.343027723306275e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25878663168.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8324900269508362, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4994651973247528, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17586894848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16901465088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23455432704.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 112194002944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151439179776.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25878663168.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6816529631614685, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.499603807926178, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17586894848.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31640592.473561075, + "step": 162, + "teaching_rms": 15457191.931064062, + "traffic_rms": 31640592.473561075 + }, + { + "batch_loss": 2.872565269470215, + "innovation_rms": 17287498.942996245, + "instruction_rms": 0.0059024442096215525, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 141668040704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.519014594223943e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23332683776.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8728030323982239, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5235716700553894, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19845404672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19506978816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26055135232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 128627638272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164336353280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23332683776.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6937556862831116, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5235077142715454, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19845404672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 36670751.14949346, + "step": 163, + "teaching_rms": 17287498.942996245, + "traffic_rms": 36670751.14949346 + }, + { + "batch_loss": 2.568795919418335, + "innovation_rms": 20065005.49360381, + "instruction_rms": 0.005795715917667851, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 169891676160.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1163581303121141e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27681466368.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9259412288665771, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8656815886497498, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21763018752.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 22432776192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29257977856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 146739314688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 181116780544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27681466368.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6995585560798645, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8656108379364014, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21763018752.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41702587.84411298, + "step": 164, + "teaching_rms": 20065005.49360381, + "traffic_rms": 41702587.84411298 + }, + { + "batch_loss": 2.7230172157287598, + "innovation_rms": 24572346.8088311, + "instruction_rms": 0.0049272579329536405, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 197681971200.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4389426399613521e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33978716160.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8853895664215088, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5811657309532166, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23246727168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25812393984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33796175872.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 167539048448.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207997255680.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33978716160.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.697184681892395, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5810949802398682, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23246727168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 46815182.45400706, + "step": 165, + "teaching_rms": 24572346.8088311, + "traffic_rms": 46815182.45400706 + }, + { + "batch_loss": 2.69415283203125, + "innovation_rms": 32081223.176771913, + "instruction_rms": 0.00348279158277103, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 220934848512.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6134430517731233e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56989220864.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8326349854469299, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5275455713272095, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27361011712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29913653248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41012596736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 193282146304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257431044096.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56989220864.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6940776109695435, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5272722840309143, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27361011712.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52031797.83304498, + "step": 166, + "teaching_rms": 32081223.176771913, + "traffic_rms": 52031797.83304498 + }, + { + "batch_loss": 2.81046986579895, + "innovation_rms": 45731216.149083115, + "instruction_rms": 0.003788129477345103, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 230166052864.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5908471413627173e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103348371456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8063433766365051, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3976256251335144, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 37695848448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35326177280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54125240320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 229544968192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 362628251648.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103348371456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6894754767417908, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39743658900260925, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 37695848448.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 54558415.80791117, + "step": 167, + "teaching_rms": 45731216.149083115, + "traffic_rms": 54558415.80791117 + }, + { + "batch_loss": 2.8159492015838623, + "innovation_rms": 40679541.90185999, + "instruction_rms": 0.0039037946682139823, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 233677520896.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.610192128382106e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115399548928.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.810299277305603, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5098607540130615, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 49235804160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 41493155840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61669769216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 270621966336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410770014208.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115399548928.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6723639965057373, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5096845626831055, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 49235804160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 53932103.04179568, + "step": 168, + "teaching_rms": 40679541.90185999, + "traffic_rms": 53932103.04179568 + }, + { + "batch_loss": 3.24359130859375, + "innovation_rms": 50076755.62328871, + "instruction_rms": 0.003950249426354654, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 306204540928.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2811905300334492e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110038646784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8558560609817505, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.814812183380127, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 60239667200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 48624291840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71311376384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 318177705984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 475557330944.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110038646784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6936272382736206, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8150429725646973, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 60239667200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 81676875.64466968, + "step": 169, + "teaching_rms": 50076755.62328871, + "traffic_rms": 81676875.64466968 + }, + { + "batch_loss": 2.4979677200317383, + "innovation_rms": 54230245.00539085, + "instruction_rms": 0.005670979461578608, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 396635439104.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6807334286547506e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109466836992.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9249356389045715, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8691592216491699, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 71186350080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 56676081664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80517890048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 371225264128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 530475548672.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109466836992.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7379814386367798, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8690630197525024, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 71186350080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 96664635.06577267, + "step": 170, + "teaching_rms": 54230245.00539085, + "traffic_rms": 96664635.06577267 + }, + { + "batch_loss": 2.5224287509918213, + "innovation_rms": 67873080.3150974, + "instruction_rms": 0.004264726631453821, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 478918049792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.9810771145438316e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119629627392.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9652481079101562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5871254205703735, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83042484224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 66087337984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94112571392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 433049436160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 618241720320.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119629627392.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7622495889663696, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5870292782783508, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83042484224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 112230914.34528247, + "step": 171, + "teaching_rms": 67873080.3150974, + "traffic_rms": 112230914.34528247 + }, + { + "batch_loss": 2.828829526901245, + "innovation_rms": 100461004.85562463, + "instruction_rms": 0.005789575276121331, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 526193197056.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.336621893605925e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280045748224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.885712206363678, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7953590154647827, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92934086656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 78256881664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121695453184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 516655415296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 836059922432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280045748224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.761766791343689, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7950858473777771, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92934086656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 129722752.8127175, + "step": 172, + "teaching_rms": 100461004.85562463, + "traffic_rms": 129722752.8127175 + }, + { + "batch_loss": 2.299107074737549, + "innovation_rms": 43410888.35829332, + "instruction_rms": 0.005728489175624179, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 505161646080.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.311995287943168e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164282875904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8058905601501465, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7982166409492493, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109362372608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 89550954496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112940744704.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 595069960192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 784145252352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164282875904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7495996356010437, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7979434728622437, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109362372608.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 101440711.88978434, + "step": 173, + "teaching_rms": 43410888.35829332, + "traffic_rms": 101440711.88978434 + }, + { + "batch_loss": 2.768159866333008, + "innovation_rms": 35519478.9492012, + "instruction_rms": 0.0061457778034747805, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 499553370112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.641542993795527e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 196642537472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7637598514556885, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5561177134513855, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 129026629632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 99479158784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99282051072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 666352418816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 712824586240.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 196642537472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7438462376594543, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5561221837997437, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 129026629632.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 121694577.1273709, + "step": 174, + "teaching_rms": 35519478.9492012, + "traffic_rms": 121694577.1273709 + }, + { + "batch_loss": 2.7075717449188232, + "innovation_rms": 58661687.06428962, + "instruction_rms": 0.003423615886335814, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 539811348480.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0825424922220326e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257823244288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7575544118881226, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5661295652389526, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 154808958976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 109195010048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97158520832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 734306041856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 679536361472.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257823244288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7518399953842163, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5661340951919556, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 154808958976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 150991510.84024864, + "step": 175, + "teaching_rms": 58661687.06428962, + "traffic_rms": 150991510.84024864 + }, + { + "batch_loss": 2.673130512237549, + "innovation_rms": 134580668.2403547, + "instruction_rms": 0.010903054453857837, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 664834605056.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2274685984467928e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 547988537344.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7783426642417908, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39276430010795593, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 186637320192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 122309058560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 131140493312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 823442931712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 891368898560.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 547988537344.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7600864768028259, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39293479919433594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 186637320192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 198788315.7719708, + "step": 176, + "teaching_rms": 134580668.2403547, + "traffic_rms": 198788315.7719708 + }, + { + "batch_loss": 2.6449360847473145, + "innovation_rms": 142162785.85234883, + "instruction_rms": 0.010140266376372655, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 812360531968.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.994837029026444e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384151355392.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8229458928108215, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5485856533050537, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 217077645312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 138575544320.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162664775680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 931022831616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1075798736896.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384151355392.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7653326988220215, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5487561225891113, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 217077645312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 244176137.37716997, + "step": 177, + "teaching_rms": 142162785.85234883, + "traffic_rms": 244176137.37716997 + }, + { + "batch_loss": 2.3303539752960205, + "innovation_rms": 139446820.97787598, + "instruction_rms": 0.002952355595794388, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 959763185664.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.640563661850412e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303906422784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8429561853408813, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3604130744934082, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 247468294144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 157145776128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 185702350848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1050203455488.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1191806369792.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303906422784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7688227891921997, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3602433204650879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 247468294144.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 265956997.88001856, + "step": 178, + "teaching_rms": 139446820.97787598, + "traffic_rms": 265956997.88001856 + }, + { + "batch_loss": 2.6437110900878906, + "innovation_rms": 151394241.04632127, + "instruction_rms": 0.0038468663929540022, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1124179771392.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.7858414548453315e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 295112310784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7778967618942261, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6505941152572632, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 276979515392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 177968250880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 208224681984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1180483715072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1302802595840.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 295112310784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7654041647911072, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6503211259841919, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 276979515392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 299620944.29931897, + "step": 179, + "teaching_rms": 151394241.04632127, + "traffic_rms": 299620944.29931897 + }, + { + "batch_loss": 2.521085500717163, + "innovation_rms": 179558498.376386, + "instruction_rms": 0.0038907850413059794, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1330022187008.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.06579921455278e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286919557120.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7199149131774902, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5798187255859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 305671471104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 201812082688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238438383616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1326932033536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1464482791424.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286919557120.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.757225751876831, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.579545795917511, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 305671471104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 348095269.2880744, + "step": 180, + "teaching_rms": 179558498.376386, + "traffic_rms": 348095269.2880744 + }, + { + "batch_loss": 2.5648040771484375, + "innovation_rms": 231822952.99987444, + "instruction_rms": 0.002968797592626092, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1590129065984.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.913492114537985e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 326766886912.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6852868795394897, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42511555552482605, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 338348146688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 230574243840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287621644288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1503791022080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1768590409728.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 326766886912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7493706345558167, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4250198304653168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 338348146688.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 410741915.016478, + "step": 181, + "teaching_rms": 231822952.99987444, + "traffic_rms": 410741915.016478 + }, + { + "batch_loss": 2.8167917728424072, + "innovation_rms": 295273081.7683269, + "instruction_rms": 0.0028049389155944755, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1877997649920.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.096180472757621e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 514040332288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6788548231124878, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42952460050582886, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 369048682496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 266356916224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 357826723840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1729668972544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2258778980352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 514040332288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7712339162826538, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42951899766921997, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 369048682496.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 470057426.83713585, + "step": 182, + "teaching_rms": 295273081.7683269, + "traffic_rms": 470057426.83713585 + }, + { + "batch_loss": 2.6939258575439453, + "innovation_rms": 395870755.7190944, + "instruction_rms": 0.0030732346493553622, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2060179865600.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.088241094988928e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 477202874368.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6923514604568481, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3264474868774414, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 389446696960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 312994234368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 466373050368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2041594642432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3119256961024.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 477202874368.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.784034013748169, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32651710510253906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 389446696960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 485200344.8284129, + "step": 183, + "teaching_rms": 395870755.7190944, + "traffic_rms": 485200344.8284129 + }, + { + "batch_loss": 2.4462578296661377, + "innovation_rms": 305019953.8339991, + "instruction_rms": 0.005295114389071895, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1878007087104.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1282033719512344e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 600741249024.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7222168445587158, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4767082631587982, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 409521094656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 363293638656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 502994010112.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2374637715456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3330430730240.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 600741249024.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7858151793479919, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4767778515815735, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 409521094656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 447246310.6555899, + "step": 184, + "teaching_rms": 305019953.8339991, + "traffic_rms": 447246310.6555899 + }, + { + "batch_loss": 2.417820930480957, + "innovation_rms": 537015810.7071605, + "instruction_rms": 0.004737240324933426, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2162494406656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2723915543698507e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 726958931968.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7323632836341858, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3829520642757416, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 434810683392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 428336873472.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 650432413696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2819546677248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4449089093632.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 726958931968.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7866396903991699, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3825291097164154, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 434810683392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 693891639.3530699, + "step": 185, + "teaching_rms": 537015810.7071605, + "traffic_rms": 693891639.3530699 + }, + { + "batch_loss": 2.4410815238952637, + "innovation_rms": 613781301.123877, + "instruction_rms": 0.002376008262547921, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2683013955584.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5309063875466375e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 841096626176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7025290131568909, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3694726824760437, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 454706364416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 508365570048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 800287031296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3378404655104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5588578729984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 841096626176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7846324443817139, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36970844864845276, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 454706364416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 845784231.062739, + "step": 186, + "teaching_rms": 613781301.123877, + "traffic_rms": 845784231.062739 + }, + { + "batch_loss": 2.387233018875122, + "innovation_rms": 850735083.7033907, + "instruction_rms": 0.002811208112138846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2835530907648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6125864085772023e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1138541985792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.706950306892395, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40852969884872437, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 512724566016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 614287933440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1059223764992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4144104472576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7656997650432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1138541985792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7695409655570984, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40876543521881104, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 512724566016.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 929504927.258589, + "step": 187, + "teaching_rms": 850735083.7033907, + "traffic_rms": 929504927.258589 + }, + { + "batch_loss": 2.604443073272705, + "innovation_rms": 865245967.0254415, + "instruction_rms": 0.0034416198039165977, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3819937202176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5214396188055505e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1378318024704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7113418579101562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5047478675842285, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 590001799168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 738814001152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1245260808192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5046105276416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9020010135552.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1378318024704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7544699907302856, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.504920244216919, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 590001799168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1206348730.1919563, + "step": 188, + "teaching_rms": 865245967.0254415, + "traffic_rms": 1206348730.1919563 + }, + { + "batch_loss": 2.2288215160369873, + "innovation_rms": 1188039122.9058151, + "instruction_rms": 0.002967471486146643, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4594779815936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1980785853353914e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2750853152768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7143526077270508, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44348353147506714, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 710283952128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 894857510912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1560434966528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6194290229248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11481848479744.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2750853152768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7653575539588928, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44344860315322876, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 710283952128.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1524972239.0199838, + "step": 189, + "teaching_rms": 1188039122.9058151, + "traffic_rms": 1524972239.0199838 + }, + { + "batch_loss": 2.3547074794769287, + "innovation_rms": 552698877.0717664, + "instruction_rms": 0.0036378007327497216, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4038552453120.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.324004432911869e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2207910985728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7165927886962891, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4057435393333435, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 902799687680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1040253452288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1453959675904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7272636350464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10783461212160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2207910985728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7783669829368591, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4057086110115051, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 902799687680.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1288721734.4648309, + "step": 190, + "teaching_rms": 552698877.0717664, + "traffic_rms": 1288721734.4648309 + }, + { + "batch_loss": 2.5656588077545166, + "innovation_rms": 702147947.1971718, + "instruction_rms": 0.0037556186401229543, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3619371352064.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.783234122862663e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2546047909888.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7212170958518982, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.284549355506897, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1157404426240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1179667267584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1394138021888.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8309031567360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10363950071808.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2546047909888.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8021175861358643, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28458601236343384, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1157404426240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1502954254.139587, + "step": 191, + "teaching_rms": 702147947.1971718, + "traffic_rms": 1502954254.139587 + }, + { + "batch_loss": 2.3503150939941406, + "innovation_rms": 1784830045.098254, + "instruction_rms": 0.013372060401006651, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4802123137024.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.217658432259893e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12874991796224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7287370562553406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36070576310157776, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1610130259968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1370624884736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1909576433664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9700526522368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13914947452928.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12874991796224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8237875699996948, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3606222867965698, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1610130259968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2433369390.6713853, + "step": 192, + "teaching_rms": 1784830045.098254, + "traffic_rms": 2433369390.6713853 + }, + { + "batch_loss": 2.7587907314300537, + "innovation_rms": 1147132809.2102509, + "instruction_rms": 0.018450877122490272, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6725767266304.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.795326599962171e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9948838756352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7323401570320129, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30708956718444824, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2605014056960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1568956350464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1983314657280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11085882064896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13853553328128.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9948838756352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.822829008102417, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3069200813770294, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2605014056960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2568352271.096742, + "step": 193, + "teaching_rms": 1147132809.2102509, + "traffic_rms": 2568352271.096742 + }, + { + "batch_loss": 2.5302414894104004, + "innovation_rms": 1886309674.8306632, + "instruction_rms": 0.005397850588063651, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9795539042304.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0092053703605912e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8275717783552.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7268118262290955, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28057217597961426, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432585887744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1806848229376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2378918789120.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12685736411136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15998544510976.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8275717783552.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8126923441886902, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28073108196258545, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432585887744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3256338059.9511323, + "step": 194, + "teaching_rms": 1886309674.8306632, + "traffic_rms": 3256338059.9511323 + }, + { + "batch_loss": 2.41984486579895, + "innovation_rms": 2207917640.226632, + "instruction_rms": 0.0025144507678817886, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13176304304128.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.65166103517232e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6477910114304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7200918197631836, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.519202470779419, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4080377004032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2090741661696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2838934323200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14553040551936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18673041408000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6477910114304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8073276877403259, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5192952156066895, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4080377004032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3713209702.432145, + "step": 195, + "teaching_rms": 2207917640.226632, + "traffic_rms": 3713209702.432145 + }, + { + "batch_loss": 2.5647807121276855, + "innovation_rms": 2876266513.733317, + "instruction_rms": 0.0022158616724418183, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15800435998720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.774064707552224e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5871948005376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.718661904335022, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4223678410053253, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4456651948032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2443886460928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3531448778752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16894306287616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23412653162496.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5871948005376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8057799935340881, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4224605858325958, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4456651948032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4295741548.299857, + "step": 196, + "teaching_rms": 2876266513.733317, + "traffic_rms": 4295741548.299857 + }, + { + "batch_loss": 2.59849214553833, + "innovation_rms": 3787188211.637263, + "instruction_rms": 0.0033652702034571327, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 16536321392640.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.136087132853394e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5418260627456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7243657112121582, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4167233109474182, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4885984051200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2898223169536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4543366037504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20026004340736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31316988919808.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5418260627456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8048396110534668, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41683194041252136, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4885984051200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4573850572.217027, + "step": 197, + "teaching_rms": 3787188211.637263, + "traffic_rms": 4573850572.217027 + }, + { + "batch_loss": 2.750023603439331, + "innovation_rms": 4374704793.154048, + "instruction_rms": 0.003425061449661187, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 18729614180352.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.820260729646217e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10517918777344.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7331417798995972, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33076566457748413, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5102527578112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3463615348736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5653921267712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23984521347072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39585165869056.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10517918777344.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8024449348449707, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3308742940425873, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5102527578112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5739893980.385812, + "step": 198, + "teaching_rms": 4374704793.154048, + "traffic_rms": 5739893980.385812 + }, + { + "batch_loss": 2.700495719909668, + "innovation_rms": 5458103648.852044, + "instruction_rms": 0.003935297457523731, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20516899389440.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.004674976904303e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8518299025408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7515842318534851, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.483969509601593, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5110211018752.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4175738437632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7121231937536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 29035004428288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50504839200768.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8518299025408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7994195222854614, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48366856575012207, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5110211018752.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7297014418.010569, + "step": 199, + "teaching_rms": 5458103648.852044, + "traffic_rms": 7297014418.010569 + }, + { + "batch_loss": 2.819509744644165, + "innovation_rms": 7340082227.269838, + "instruction_rms": 0.004456410926522667, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 22967455055872.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.990201343324474e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8293204361216.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7471051216125488, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5505599975585938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5717564063744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5101956628480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9262179811328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 35847051149312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68120454627328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8293204361216.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7952286005020142, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5504636168479919, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5717564063744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8366655466.334008, + "step": 200, + "teaching_rms": 7340082227.269838, + "traffic_rms": 8366655466.334008 + }, + { + "batch_loss": 2.8091506958007812, + "innovation_rms": 7478652317.526306, + "instruction_rms": 0.004425276534136918, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24297745678336.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.354915196888374e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22109426286592.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7403864860534668, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44552966952323914, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6612322353152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6188284510208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10863276720128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 43853637746688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80065874362368.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22109426286592.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7908715009689331, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4457092881202698, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6612322353152.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9021598308.3086, + "step": 201, + "teaching_rms": 7478652317.526306, + "traffic_rms": 9021598308.3086 + }, + { + "batch_loss": 2.653174638748169, + "innovation_rms": 7008671780.068236, + "instruction_rms": 0.0055367964713973206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36783274328064.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.948803745878161e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24083081199616.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7400909662246704, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3790639042854309, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9020630892544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7365245009920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11769607094272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 52273963073536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84203253268480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24083081199616.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7915959358215332, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37924349308013916, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9020630892544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12674554382.680796, + "step": 202, + "teaching_rms": 7008671780.068236, + "traffic_rms": 12674554382.680796 + }, + { + "batch_loss": 2.725430488586426, + "innovation_rms": 7370542926.268285, + "instruction_rms": 0.004336172018675045, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56530112086016.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2586536668448198e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27651087532032.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7310705780982971, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38259920477867126, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11785739436032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8613419548672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12481743290368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 60916305494016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86423432593408.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27651087532032.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7961867451667786, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38259363174438477, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11785739436032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15049847860.94992, + "step": 203, + "teaching_rms": 7370542926.268285, + "traffic_rms": 15049847860.94992 + }, + { + "batch_loss": 2.591639995574951, + "innovation_rms": 8424922097.819022, + "instruction_rms": 0.0036358891362698463, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 81048071307264.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.688196790196864e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31140855414784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.725982666015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38853925466537476, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14899824558080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9948769550336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13353497919488.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 69874147655680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89578438393856.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31140855414784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7978097200393677, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3886605203151703, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14899824558080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17605379900.705063, + "step": 204, + "teaching_rms": 8424922097.819022, + "traffic_rms": 17605379900.705063 + }, + { + "batch_loss": 2.8301310539245605, + "innovation_rms": 10204885878.341055, + "instruction_rms": 0.005859470915319631, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 108809095741440.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.602296478119917e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37808475471872.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7288800477981567, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5041475296020508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18680672944128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11418607812608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14698378428416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 79490826895360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 96166817562624.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37808475471872.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7926801443099976, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5041823983192444, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18680672944128.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20375692264.283356, + "step": 205, + "teaching_rms": 10204885878.341055, + "traffic_rms": 20375692264.283356 + }, + { + "batch_loss": 2.6287221908569336, + "innovation_rms": 12008498501.6511, + "instruction_rms": 0.005208117074841458, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 137635288842240.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.72095521012934e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44913076994048.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7454500794410706, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6048339605331421, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23171979804672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13061602672640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16429943357440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 90066403721216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105755801812992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44913076994048.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7728002667427063, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6048635244369507, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23171979804672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23445726191.430565, + "step": 206, + "teaching_rms": 12008498501.6511, + "traffic_rms": 23445726191.430565 + }, + { + "batch_loss": 3.310581922531128, + "innovation_rms": 14168327060.864243, + "instruction_rms": 0.003473178503112106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 165197880754176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.681920253324201e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48774688800768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7684115171432495, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6589593887329102, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28049447845888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 14931593265152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18699905925120.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 101980089352192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119136889864192.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48774688800768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7445924282073975, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6589288711547852, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28049447845888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 27220262424.51282, + "step": 207, + "teaching_rms": 14168327060.864243, + "traffic_rms": 27220262424.51282 + }, + { + "batch_loss": 2.406362533569336, + "innovation_rms": 17943967823.72172, + "instruction_rms": 0.0037836277177959315, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 188767621087232.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.01022433988403e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52485867700224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7700548768043518, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6759541630744934, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33298034196480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17152970063872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22213763792896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 116237082296320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142569929441280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52485867700224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7424826622009277, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.676176905632019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33298034196480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31503047436.132935, + "step": 208, + "teaching_rms": 17943967823.72172, + "traffic_rms": 31503047436.132935 + }, + { + "batch_loss": 2.9623093605041504, + "innovation_rms": 24161703439.295746, + "instruction_rms": 0.0038726518632135804, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 203472632807424.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.071148534352828e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50817730084864.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7434528470039368, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5133477449417114, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38357287567360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19961162498048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28081928536064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 134947578839040.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 187104948649984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50817730084864.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.755883514881134, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5135051012039185, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38357287567360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 36543847486.567856, + "step": 209, + "teaching_rms": 24161703439.295746, + "traffic_rms": 36543847486.567856 + }, + { + "batch_loss": 2.8970119953155518, + "innovation_rms": 32694946491.66783, + "instruction_rms": 0.004575950447594616, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 198734881226752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0634598133637437e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60986602029056.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7340909838676453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6396793723106384, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44455948189696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 23646363975680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36852006387712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 161443206922240.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 264956247277568.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60986602029056.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7733847498893738, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6395701169967651, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44455948189696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 38511489559.64006, + "step": 210, + "teaching_rms": 32694946491.66783, + "traffic_rms": 38511489559.64006 + }, + { + "batch_loss": 2.917597532272339, + "innovation_rms": 38417103621.31518, + "instruction_rms": 0.004545626221620703, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 197780274413568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.097567577522918e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91987386564608.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7408406734466553, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5880359411239624, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 53654686007296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 28290255421440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46438910263296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 196415967985664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 349727677743104.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91987386564608.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.78750079870224, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5880063772201538, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 53654686007296.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 45130974073.00624, + "step": 211, + "teaching_rms": 38417103621.31518, + "traffic_rms": 45130974073.00624 + }, + { + "batch_loss": 2.3079793453216553, + "innovation_rms": 48745476565.422264, + "instruction_rms": 0.004442736182965311, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 211466540023808.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2335026676197518e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116148926414848.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7479676008224487, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43336808681488037, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 65269577809920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34182682116096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58924271140864.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 242586916749312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 461709554745344.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116148926414848.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7925726771354675, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43333855271339417, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 65269577809920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 59570642546.547386, + "step": 212, + "teaching_rms": 48745476565.422264, + "traffic_rms": 59570642546.547386 + }, + { + "batch_loss": 2.6065492630004883, + "innovation_rms": 61919515222.1426, + "instruction_rms": 0.007728519866181343, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 211038133813248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3748198387971122e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125123696787456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7493404150009155, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27521073818206787, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 77781950005248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 41630312693760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74476293193728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 303486616993792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 608997103108096.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125123696787456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7855510711669922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2752383053302765, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 77781950005248.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 71433551332.1165, + "step": 213, + "teaching_rms": 61919515222.1426, + "traffic_rms": 71433551332.1165 + }, + { + "batch_loss": 2.648090362548828, + "innovation_rms": 68399377135.45827, + "instruction_rms": 0.010646286802676758, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 220458590928896.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6067522572901788e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 143908080189440.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7506697773933411, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41451117396354675, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92172758024192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 50364892053504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87345776820224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 376828686499840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 733420627951616.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 143908080189440.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7717603445053101, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41453874111175537, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92172758024192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 88413067246.9395, + "step": 214, + "teaching_rms": 68399377135.45827, + "traffic_rms": 88413067246.9395 + }, + { + "batch_loss": 2.6899123191833496, + "innovation_rms": 91627191836.94122, + "instruction_rms": 0.005030675893177306, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 229534712463360.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.069325172865761e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230200012963840.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7495959401130676, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37820175290107727, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 102833319837696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 61231801040896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108669073096704.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 469863583711232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 930348972113920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230200012963840.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7729347944259644, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37838244438171387, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 102833319837696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 116439691745.13644, + "step": 215, + "teaching_rms": 91627191836.94122, + "traffic_rms": 116439691745.13644 + }, + { + "batch_loss": 2.4526922702789307, + "innovation_rms": 118031984815.88239, + "instruction_rms": 0.003420080067080699, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 247412648050688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6586942188022034e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171467979358208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7500817179679871, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5793404579162598, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 119980121128960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 75007224971264.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137754197360640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 589912013275136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1200484295639040.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171467979358208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7843001484870911, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5793532729148865, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 119980121128960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 139084361129.51794, + "step": 216, + "teaching_rms": 118031984815.88239, + "traffic_rms": 139084361129.51794 + }, + { + "batch_loss": 2.487031936645508, + "innovation_rms": 147928819920.1325, + "instruction_rms": 0.004300500239234131, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 276242330812416.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9390548039665534e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309872964927488.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7507343292236328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34346139430999756, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150967412588544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 92359480049664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 173522567561216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 743379784695808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1534677579988992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309872964927488.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7897621989250183, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3434741795063019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150967412588544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 173400948512.61935, + "step": 217, + "teaching_rms": 147928819920.1325, + "traffic_rms": 173400948512.61935 + }, + { + "batch_loss": 2.704700231552124, + "innovation_rms": 190093542878.07742, + "instruction_rms": 0.0035161578742184665, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 320362315251712.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.310024205463827e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 494469317132288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7562136650085449, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33246558904647827, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 200414347657216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 114260793360384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219013133107200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 941063640449024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1976838423314432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 494469317132288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7874566912651062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33277052640914917, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 200414347657216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 221673061690.87836, + "step": 218, + "teaching_rms": 190093542878.07742, + "traffic_rms": 221673061690.87836 + }, + { + "batch_loss": 2.307468891143799, + "innovation_rms": 246910479414.83594, + "instruction_rms": 0.004161785210674477, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 296917128970240.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.816839388543377e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 708584442691584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7592132687568665, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40762007236480713, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 271272785215488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 141754825179136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 274940368519168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1198955287281664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2578916870979584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 708584442691584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.781217098236084, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4077376127243042, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 271272785215488.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 273795571846.72592, + "step": 219, + "teaching_rms": 246910479414.83594, + "traffic_rms": 273795571846.72592 + }, + { + "batch_loss": 2.484511375427246, + "innovation_rms": 304906963981.69916, + "instruction_rms": 0.005772476934369502, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 418512149413888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.030361635370576e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1032969833676800.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7662621736526489, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33628910779953003, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 374569768583168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 176294616629248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 345397880946688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1526086303219712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3271310159380480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1032969833676800.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7688987851142883, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33610692620277405, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 374569768583168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 357319688230.1643, + "step": 220, + "teaching_rms": 304906963981.69916, + "traffic_rms": 357319688230.1643 + }, + { + "batch_loss": 2.3402576446533203, + "innovation_rms": 401169246043.6957, + "instruction_rms": 0.004962158521820311, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 442098331418624.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.828369994947132e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818585266028544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7686542272567749, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3928506374359131, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 454083303440384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 221230460305408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 449358470316032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1950055275692032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4239689187852288.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818585266028544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7527469396591187, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3928781747817993, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 454083303440384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 470472727663.74286, + "step": 221, + "teaching_rms": 401169246043.6957, + "traffic_rms": 470472727663.74286 + }, + { + "batch_loss": 2.283388137817383, + "innovation_rms": 471948208031.3073, + "instruction_rms": 0.006927568611305808, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 385162835656704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.903815307551027e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1416929306738688.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7708644866943359, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2003263682126999, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 514801088331776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 274449987272704.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 532195303227392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2470962599559168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5209072701800448.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1416929306738688.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7536391615867615, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20047083497047424, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 514801088331776.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 604986583897.1133, + "step": 222, + "teaching_rms": 471948208031.3073, + "traffic_rms": 604986583897.1133 + }, + { + "batch_loss": 2.545536756515503, + "innovation_rms": 546751779181.35175, + "instruction_rms": 0.0070258755389610095, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 580235451957248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.836655382778794e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1086532572151808.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7694854140281677, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2548127770423889, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 623454365679616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 339840897908736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 653909039251456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3085629059497984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6146665136259072.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1086532572151808.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7682890892028809, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25484633445739746, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 623454365679616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 694741517191.5022, + "step": 223, + "teaching_rms": 546751779181.35175, + "traffic_rms": 694741517191.5022 + }, + { + "batch_loss": 2.265681028366089, + "innovation_rms": 773748650129.5637, + "instruction_rms": 0.008168602148467808, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 901771870339072.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.735193896251995e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2062611873005568.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7651926279067993, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37583470344543457, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 829715539558400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 422548445593600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 827075611066368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3876642313207808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7910133610840064.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2062611873005568.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7785825729370117, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3759917616844177, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 829715539558400.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 916951628515.1971, + "step": 224, + "teaching_rms": 773748650129.5637, + "traffic_rms": 916951628515.1971 + }, + { + "batch_loss": 2.707423210144043, + "innovation_rms": 985240759122.5924, + "instruction_rms": 0.15204283129863563, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 887936304283648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.898555603956804e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2709496493244416.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7674070000648499, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40815290808677673, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1100665162039296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 505014669803520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 824662107881472.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4898332354805760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0216902563463168e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2709496493244416.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7784802913665771, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4080438017845154, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1100665162039296.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1162303319515.5493, + "step": 225, + "teaching_rms": 985240759122.5924, + "traffic_rms": 1162303319515.5493 + }, + { + "batch_loss": 2.631718873977661, + "innovation_rms": 277855971931.93274, + "instruction_rms": 0.0018064147381540887, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 750339779198976.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.092573418758594e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2466476237783040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7714658975601196, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4323897957801819, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1347312852926464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 579562752704512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 745480963227648.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5805221678678016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9068894312464384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2466476237783040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7595797777175903, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4324236512184143, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1347312852926464.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1667787445260.4, + "step": 226, + "teaching_rms": 277855971931.93274, + "traffic_rms": 1667787445260.4 + }, + { + "batch_loss": 2.396836519241333, + "innovation_rms": 373452082854.3004, + "instruction_rms": 0.0017956001126870989, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 575106187264000.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.869725397458625e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2311002985070592.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7788482308387756, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3569227159023285, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1578413097746432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 643932903964672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 643701244166144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6614527198101504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8093055194234880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2311002985070592.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7399919629096985, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3568115830421448, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1578413097746432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1904007977684.3818, + "step": 227, + "teaching_rms": 373452082854.3004, + "traffic_rms": 1904007977684.3818 + }, + { + "batch_loss": 2.482481002807617, + "innovation_rms": 263385793908.45956, + "instruction_rms": 0.001517916838672304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 614677331574784.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4854685679921313e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2328407534731264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7786386609077454, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49258923530578613, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1773747405062144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 697924065427456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 539911916617728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7313035446190080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6985080870273024.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2328407534731264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7421091794967651, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4927205443382263, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1773747405062144.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2072308222913.1746, + "step": 228, + "teaching_rms": 263385793908.45956, + "traffic_rms": 2072308222913.1746 + }, + { + "batch_loss": 2.6279940605163574, + "innovation_rms": 206176148861.29404, + "instruction_rms": 0.002569791625514382, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 795875894886400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.180394187955389e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2266961048240128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7771221399307251, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5791332125663757, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1928052560887808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 740987353694208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 430632748449792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7901670377783296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5886350926544896.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2266961048240128.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7327043414115906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5790993571281433, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1928052560887808.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2215073363154.0234, + "step": 229, + "teaching_rms": 206176148861.29404, + "traffic_rms": 2215073363154.0234 + }, + { + "batch_loss": 2.6119182109832764, + "innovation_rms": 264557496486.17603, + "instruction_rms": 0.003972686379897573, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 995643011104768.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6756724787938805e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2017256078835712.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7754343748092651, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3637765049934387, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2035139920003072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 776684840157184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 356974730412032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8391970422521856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4903001521127424.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2017256078835712.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7335575819015503, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3637426495552063, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2035139920003072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2314023775184.2383, + "step": 230, + "teaching_rms": 264557496486.17603, + "traffic_rms": 2314023775184.2383 + }, + { + "batch_loss": 2.355194568634033, + "innovation_rms": 505373267354.4954, + "instruction_rms": 0.004056618074972019, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1232482473082880.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9165040420302943e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1540808511586304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7736788988113403, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28683942556381226, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2026834795429888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 814337274937344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 376524414910464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824234184802304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4322638159675392.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1540808511586304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7331982851028442, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2866171896457672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2026834795429888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2322442279052.239, + "step": 231, + "teaching_rms": 505373267354.4954, + "traffic_rms": 2322442279052.239 + }, + { + "batch_loss": 2.5967702865600586, + "innovation_rms": 919916021806.2888, + "instruction_rms": 0.004575994365856686, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859278725971968.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.172708064203764e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2023748190339072.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7745083570480347, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2589944899082184, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1824459962974208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 868789407186944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 544521054060544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9289689991217152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4654552695439360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2023748190339072.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7318764328956604, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25896066427230835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1824459962974208.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2475434465281.7036, + "step": 232, + "teaching_rms": 919916021806.2888, + "traffic_rms": 2475434465281.7036 + }, + { + "batch_loss": 2.355726718902588, + "innovation_rms": 1263007167921.9336, + "instruction_rms": 0.0035295161744182623, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4042424024301568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6664045964998715e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3028167765786624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7781832218170166, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34230777621269226, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1521643159552000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 949808462299136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 810190618230784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9886121362194432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5964310488547328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3028167765786624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7318496108055115, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3422701358795166, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1521643159552000.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2651569215644.501, + "step": 233, + "teaching_rms": 1263007167921.9336, + "traffic_rms": 2651569215644.501 + }, + { + "batch_loss": 2.556157350540161, + "innovation_rms": 1480671778752.2947, + "instruction_rms": 0.0020868352918967247, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6703887213920256.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.41069960372906e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2914093165969408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7804502248764038, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3013184666633606, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230233789267968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063943996964864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1141355480875008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0663381593751552e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7772599631216640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2914093165969408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7347392439842224, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30101850628852844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230233789267968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2885747805735.3115, + "step": 234, + "teaching_rms": 1480671778752.2947, + "traffic_rms": 2885747805735.3115 + }, + { + "batch_loss": 2.46859073638916, + "innovation_rms": 1699178385809.8103, + "instruction_rms": 0.002149647299454771, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9019672913510400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.150921498086802e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3083635657801728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7813762426376343, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30286872386932373, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1221365084454912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1218466115747840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1545221590482944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1662862359461888e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9994805509619712.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3083635657801728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7370222210884094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3025687634944916, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1221365084454912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3161304226877.3438, + "step": 235, + "teaching_rms": 1699178385809.8103, + "traffic_rms": 3161304226877.3438 + }, + { + "batch_loss": 2.46981143951416, + "innovation_rms": 2081644842222.6975, + "instruction_rms": 0.0017539141336932853, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0472936301395968e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.497105627229025e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6107231430901760.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7771000266075134, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30446070432662964, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1485289214181376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1427846979387392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2093809173266432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.299248008134656e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3296172923879424e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6107231430901760.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7360103726387024, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3045779764652252, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1485289214181376.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3311254114933.46, + "step": 236, + "teaching_rms": 2081644842222.6975, + "traffic_rms": 3311254114933.46 + }, + { + "batch_loss": 2.3470146656036377, + "innovation_rms": 2492202010616.4795, + "instruction_rms": 0.002435418749988361, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.055535994503168e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4120920293124125e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4716209098457088.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7742604613304138, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39154234528541565, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865235308740608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1703874863824896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2760278307504128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4765064624013312e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7725841131700224e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4716209098457088.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7371050119400024, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39165961742401123, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865235308740608.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3066531064101.9014, + "step": 237, + "teaching_rms": 2492202010616.4795, + "traffic_rms": 3066531064101.9014 + }, + { + "batch_loss": 2.7309703826904297, + "innovation_rms": 2822611544814.047, + "instruction_rms": 0.002432404127649381, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4541786454163456e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.532022532930947e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4632954378649600.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7738653421401978, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46075111627578735, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2328530746605568.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2043494168264704.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3396193312833536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6990232419237888e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.225167795224576e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4632954378649600.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.738079309463501, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46097201108932495, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2328530746605568.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4111385151069.9297, + "step": 238, + "teaching_rms": 2822611544814.047, + "traffic_rms": 4111385151069.9297 + }, + { + "batch_loss": 2.572516679763794, + "innovation_rms": 3262027446176.384, + "instruction_rms": 0.0023408367436290557, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.923974623657984e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.429074731914157e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5206228359708672.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7758561968803406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33398178219795227, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2839115385012224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2455454814830592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4119605660352512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9720067461677056e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.729835901432627e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5206228359708672.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7393243312835693, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3342026472091675, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2839115385012224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4929468711215.347, + "step": 239, + "teaching_rms": 3262027446176.384, + "traffic_rms": 4929468711215.347 + }, + { + "batch_loss": 2.4935126304626465, + "innovation_rms": 4515194951261.788, + "instruction_rms": 0.007822943022755316, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.202619002172211e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.270107983228908e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7730118109691904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7739513516426086, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3612111806869507, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3286173430579200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2995517961273344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5400630390685696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.342473874748211e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.704671715301786e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7730118109691904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7354758381843567, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3610481321811676, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3286173430579200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5727501345917.6455, + "step": 240, + "teaching_rms": 4515194951261.788, + "traffic_rms": 5727501345917.6455 + }, + { + "batch_loss": 2.5731537342071533, + "innovation_rms": 3637616713239.794, + "instruction_rms": 0.004439128886435068, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.255234499031859e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0220189244894865e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9836936470462464.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7788320779800415, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3164953887462616, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4168858734690304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3588097820327936.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5925798590545920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.742162887068877e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.996890982200115e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9836936470462464.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7290318608283997, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3164560794830322, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4168858734690304.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5067107691211.877, + "step": 241, + "teaching_rms": 3637616713239.794, + "traffic_rms": 5067107691211.877 + }, + { + "batch_loss": 2.536705732345581, + "innovation_rms": 5909965110410.219, + "instruction_rms": 0.004348703542714681, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.800803793548083e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3574288165851067e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5711506838585344e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.781929075717926, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3074256479740143, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5110273421606912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4331001534742528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7429037144145920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.253052253405184e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.108893233866342e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5711506838585344e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7211084961891174, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3074312210083008, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5110273421606912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7947202433696.706, + "step": 242, + "teaching_rms": 5909965110410.219, + "traffic_rms": 7947202433696.706 + }, + { + "batch_loss": 2.4460833072662354, + "innovation_rms": 6960350698681.133, + "instruction_rms": 0.002884438322844351, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.242133158048563e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8091013948952624e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7222231520182272e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.776141345500946, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42963317036628723, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5652497473470464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5251039364120576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9200377220038656.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.889049077861581e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.359968674060698e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7222231520182272e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7127447724342346, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4294123649597168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5652497473470464.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10145502839514.521, + "step": 243, + "teaching_rms": 6960350698681.133, + "traffic_rms": 10145502839514.521 + }, + { + "batch_loss": 2.5950448513031006, + "innovation_rms": 7982520665538.923, + "instruction_rms": 0.01803258119055238, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499235698592973e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2011432719396316e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.028621154097562e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7730055451393127, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24842065572738647, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7028344612716544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6370261133688832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1192218769424384e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.660445390569472e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.713961409091994e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.028621154097562e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7279645800590515, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24840843677520752, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7028344612716544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9334984272538.145, + "step": 244, + "teaching_rms": 7982520665538.923, + "traffic_rms": 9334984272538.145 + }, + { + "batch_loss": 2.5697789192199707, + "innovation_rms": 8840807362761.496, + "instruction_rms": 0.013605203762016157, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.494956005687296e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2111632691702134e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.235274981526733e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7711538672447205, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3932304084300995, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734803838894080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7664167276249088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2939059278118912e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.547730228844954e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.872848382754816e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.235274981526733e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7326834797859192, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39293816685676575, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734803838894080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13606516753257.48, + "step": 245, + "teaching_rms": 8840807362761.496, + "traffic_rms": 13606516753257.48 + }, + { + "batch_loss": 2.395833969116211, + "innovation_rms": 12766551256143.781, + "instruction_rms": 0.017171621617056928, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.340409580486656e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.138083833269928e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.001021452248678e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7723332643508911, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3639097809791565, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2062758174457856e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9297151423152128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.63298414690304e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.698915037655859e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1511848947102515e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.001021452248678e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.720115065574646, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3637988865375519, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2062758174457856e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16421795203649.723, + "step": 246, + "teaching_rms": 12766551256143.781, + "traffic_rms": 16421795203649.723 + }, + { + "batch_loss": 2.4155306816101074, + "innovation_rms": 6396383370372.111, + "instruction_rms": 0.008270874443933684, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.508507915996365e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.9438391149246284e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.099558926929101e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7725947499275208, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3563154935836792, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5162316886638592e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0850674380111872e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5535232790822912e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.79466813955113e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0957532736939622e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.099558926929101e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6970468163490295, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35632771253585815, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5162316886638592e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13620205791335.53, + "step": 247, + "teaching_rms": 6396383370372.111, + "traffic_rms": 13620205791335.53 + }, + { + "batch_loss": 2.4134368896484375, + "innovation_rms": 10200543593245.146, + "instruction_rms": 0.013302389565926314, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.904839095628595e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.265897406488935e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.075293275213005e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7752251625061035, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4240496754646301, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7864612598775808e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2482102675111936e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6314279728775168e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.930121771646976e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1354532884984627e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.075293275213005e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6835389733314514, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42416054010391235, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7864612598775808e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16723701931032.65, + "step": 248, + "teaching_rms": 10200543593245.146, + "traffic_rms": 16723701931032.65 + }, + { + "batch_loss": 2.467437267303467, + "innovation_rms": 18204928642072.062, + "instruction_rms": 0.013318268200733518, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.182822318440448e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.400746176126948e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.45492218602455e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7734317183494568, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36472365260124207, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0755616288997376e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4568629999763456e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.08652732465152e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.038610022310871e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4559781078643507e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.45492218602455e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6845196485519409, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3648345172405243, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0755616288997376e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 27460539624040.71, + "step": 249, + "teaching_rms": 18204928642072.062, + "traffic_rms": 27460539624040.71 + }, + { + "batch_loss": 2.4313933849334717, + "innovation_rms": 19710084455829.94, + "instruction_rms": 0.010407892052422506, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.436750100928922e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5033166984039473e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.784769661403136e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7702292203903198, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28429552912712097, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4389686902390784e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7073819999010816e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.50518999924736e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2127169154750874e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7410689316421632e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.784769661403136e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6904570460319519, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2842617332935333, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4389686902390784e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30221280568080.035, + "step": 250, + "teaching_rms": 19710084455829.94, + "traffic_rms": 30221280568080.035 + }, + { + "batch_loss": 2.353072166442871, + "innovation_rms": 27062966153863.566, + "instruction_rms": 0.012280698885850747, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1831103315142246e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1544990010663775e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.509369970445517e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7653226852416992, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4676092267036438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9535087787769856e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.029868330831053e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.224862450306253e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4414965089343898e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.287796106391716e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.509369970445517e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6863639950752258, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46786826848983765, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9535087787769856e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35330819806254.24, + "step": 251, + "teaching_rms": 27062966153863.566, + "traffic_rms": 35330819806254.24 + }, + { + "batch_loss": 2.7815051078796387, + "innovation_rms": 797354125304.6063, + "instruction_rms": 0.006698876352578485, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.172416338696602e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.5947067762246862e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.858198382287258e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7637274265289307, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.579425573348999, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4165653090861056e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.310152313110528e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.802839608046387e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.647398100098089e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0590164270330675e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.858198382287258e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6823253035545349, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5793401598930359, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4165653090861056e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6455889015564.973, + "step": 252, + "teaching_rms": 797354125304.6063, + "traffic_rms": 6455889015564.973 + }, + { + "batch_loss": 2.4498822689056396, + "innovation_rms": 907452574244.8943, + "instruction_rms": 0.003136302978002226, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.4533494235136e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.436530535130046e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.272085240741888e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7641482353210449, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5112813711166382, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.833281934000128e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.551079792069837e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4092752190898176e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.832709618044764e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8531146640706765e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.272085240741888e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6957221031188965, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5111958980560303, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.833281934000128e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7347317304585.879, + "step": 253, + "teaching_rms": 907452574244.8943, + "traffic_rms": 7347317304585.879 + }, + { + "batch_loss": 2.910223960876465, + "innovation_rms": 1002091518885.0789, + "instruction_rms": 0.009999593588419172, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.518587290733773e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2976246876569414e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.744530971800371e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.768829882144928, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4209117293357849, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.208288413515776e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.755405130235904e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.043253166912307e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.999489898297426e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6678031461240013e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.744530971800371e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6961692571640015, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42065274715423584, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.208288413515776e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8113574448747.41, + "step": 254, + "teaching_rms": 1002091518885.0789, + "traffic_rms": 8113574448747.41 + }, + { + "batch_loss": 2.5800864696502686, + "innovation_rms": 1082352645966.4801, + "instruction_rms": 0.002930269797295857, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.37077070430208e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1787702965989094e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.269684541515366e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7746244072914124, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4906191825866699, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.545752154400358e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.925786482868224e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.703814063194112e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1495921333449523e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5010228658713395e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.269684541515366e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7169312834739685, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49036023020744324, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.545752154400358e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8763420351537.297, + "step": 255, + "teaching_rms": 1082352645966.4801, + "traffic_rms": 8763420351537.297 + }, + { + "batch_loss": 2.939842462539673, + "innovation_rms": 1149280394002.799, + "instruction_rms": 0.005048339490732889, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3962092570738688e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0806507822288879e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.842279890485248e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7718848586082458, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5699291229248047, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.849423951593472e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0647827914817536e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3899640525029376e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.284684144887726e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3509205449244672e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.842279890485248e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7426603436470032, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.570188045501709, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.849423951593472e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9305309198539.117, + "step": 256, + "teaching_rms": 1149280394002.799, + "traffic_rms": 9305309198539.117 + }, + { + "batch_loss": 2.563703775405884, + "innovation_rms": 1203879684492.7275, + "instruction_rms": 0.003530607177736849, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3219620558353203e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0036119940453522e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.45757752229888e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7693012356758118, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.616066575050354, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.122680078886502e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1748512068665344e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1006831874801664e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4062669380963533e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2158284474823475e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.45757752229888e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7576264142990112, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6162241697311401, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.122680078886502e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9747379577828.14, + "step": 257, + "teaching_rms": 1203879684492.7275, + "traffic_rms": 9747379577828.14 + }, + { + "batch_loss": 2.845197916030884, + "innovation_rms": 1247115763063.5308, + "instruction_rms": 0.003289130520624741, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2831324434097766e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.473699021805388e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.111310816444416e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7661909461021423, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6026109457015991, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.401446649220301e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.258344512107315e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8349331597819904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5156915378834637e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0942455683743744e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.111310816444416e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7567599415779114, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6027315258979797, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.401446649220301e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10097446230167.611, + "step": 258, + "teaching_rms": 1247115763063.5308, + "traffic_rms": 10097446230167.611 + }, + { + "batch_loss": 2.6869654655456543, + "innovation_rms": 1279912747714.351, + "instruction_rms": 0.005953906209918674, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.227239784910684e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.109640199064672e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7996396426625024e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7662234902381897, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42159539461135864, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.681410656436224e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.317510907835187e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5916639572787200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.614173591792517e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.84820968587264e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7996396426625024e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7496383190155029, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42171597480773926, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.681410656436224e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10362992119547.74, + "step": 259, + "teaching_rms": 1279912747714.351, + "traffic_rms": 10362992119547.74 + }, + { + "batch_loss": 3.039046049118042, + "innovation_rms": 1303153892116.71, + "instruction_rms": 0.11530611653690151, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.277399489676575e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.927936133810888e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5191074112733184e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7652134299278259, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6573208570480347, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.933321225764864e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.35449293848576e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3698204407234560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7028074574905344e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.863388287788646e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5191074112733184e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.723983645439148, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6573394536972046, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.933321225764864e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10551168008093.838, + "step": 260, + "teaching_rms": 1303153892116.71, + "traffic_rms": 10551168008093.838 + }, + { + "batch_loss": 2.807569980621338, + "innovation_rms": 1317680895838.799, + "instruction_rms": 0.00413150708694184, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.234055694824243e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.906621127471967e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.266603277464371e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7685402631759644, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41625991463661194, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.15998168236032e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3713277070475264e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1683478064136192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7825778850791424e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.977049630808474e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.266603277464371e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7501400113105774, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4160394072532654, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.15998168236032e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10668787170122.03, + "step": 261, + "teaching_rms": 1317680895838.799, + "traffic_rms": 10668787170122.03 + }, + { + "batch_loss": 2.949625253677368, + "innovation_rms": 1324293779709.48, + "instruction_rms": 0.005404326368275987, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.178048961993769e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.02021373176427e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.039326922558669e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.775852382183075, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41973310708999634, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.363914460515533e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.369947304558592e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 138030719434752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8543713558082355e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.179344495928934e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.039326922558669e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7697267532348633, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4196125268936157, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.363914460515533e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10722329389813.938, + "step": 262, + "teaching_rms": 1324293779709.48, + "traffic_rms": 10722329389813.938 + }, + { + "batch_loss": 2.8665709495544434, + "innovation_rms": 1323751478206.049, + "instruction_rms": 0.0064671642438244034, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.11077498375766e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.240838631853185e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8347578449985536e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7645182609558105, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43572086095809937, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.547390309439898e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.352180098596864e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1776712274673664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.918985531004027e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.461409788638003e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8347578449985536e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.771552324295044, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43560031056404114, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.547390309439898e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10717938616843.871, + "step": 263, + "teaching_rms": 1323751478206.049, + "traffic_rms": 10717938616843.871 + }, + { + "batch_loss": 2.7025599479675293, + "innovation_rms": 1316772371450.349, + "instruction_rms": 0.005235688164364544, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.97772066080555e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.539270384548761e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6506272283099136e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.755246639251709, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2941921353340149, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.712453204069581e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.31975202177024e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3242813051371520.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.977138357399716e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.815268466176819e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6506272283099136e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.760421872138977, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29392147064208984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.712453204069581e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10661431860729.645, + "step": 264, + "teaching_rms": 1316772371450.349, + "traffic_rms": 10661431860729.645 + }, + { + "batch_loss": 2.7208292484283447, + "innovation_rms": 1304034233683.8755, + "instruction_rms": 0.005357734799491967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.797519279254405e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.887710272556319e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4848932450402304e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7490054965019226, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3951801359653473, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.860942678897459e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2742880749551616e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4546402197700608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.029475798076621e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.233741490710118e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4848932450402304e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7390521168708801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3953673839569092, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.860942678897459e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10558295677053.01, + "step": 265, + "teaching_rms": 1304034233683.8755, + "traffic_rms": 10558295677053.01 + }, + { + "batch_loss": 2.8007752895355225, + "innovation_rms": 1286175596782.0308, + "instruction_rms": 0.0070148898557989845, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0580081562039091e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0260525199937406e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3357177565609984e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7486642599105835, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5026673078536987, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.994514443816141e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.217314689528627e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5697339079524352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.076579563405312e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.710367083941069e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3357177565609984e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7152320742607117, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5026628971099854, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.994514443816141e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10413699752422.53, + "step": 266, + "teaching_rms": 1286175596782.0308, + "traffic_rms": 10413699752422.53 + }, + { + "batch_loss": 2.807339906692505, + "innovation_rms": 1263795733235.508, + "instruction_rms": 0.00941808597382605, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1319784881573069e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0634663175735479e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2014465562181632e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7541813850402832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38087302446365356, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.114659281974067e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1502623043485696e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6705244423585792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.118972952201134e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.239330289647616e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2014465562181632e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7132059335708618, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3808686137199402, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.114659281974067e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10232499214527.984, + "step": 267, + "teaching_rms": 1263795733235.508, + "traffic_rms": 10232499214527.984 + }, + { + "batch_loss": 2.7913944721221924, + "innovation_rms": 1237456914496.872, + "instruction_rms": 0.00492854575262196, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2069593400161075e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0989603796374334e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0805903641018368e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7665583491325378, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4003666639328003, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.22271851165778e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.074467513237504e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7579479647977472.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.157127036477112e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.815397260682854e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0805903641018368e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7308158874511719, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40039998292922974, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.22271851165778e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10019242700433.545, + "step": 268, + "teaching_rms": 1237456914496.872, + "traffic_rms": 10019242700433.545 + }, + { + "batch_loss": 2.8569881916046143, + "innovation_rms": 1207683791922.9883, + "instruction_rms": 0.0047981362059391856, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2784292450889892e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1308145111295576e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9718090430414848.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7708239555358887, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42912957072257996, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.319899018677453e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.991176286208e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8329124850434048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.191465471807324e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.433857491664896e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9718090430414848.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7420187592506409, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4291628897190094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.319899018677453e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9778181335882.113, + "step": 269, + "teaching_rms": 1207683791922.9883, + "traffic_rms": 9778181335882.113 + }, + { + "batch_loss": 2.813779592514038, + "innovation_rms": 1174966171479.231, + "instruction_rms": 0.0074027822216367264, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3447684165898076e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.15766109779152e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738961367236608.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7609487771987915, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44913744926452637, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.407288718249165e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.901546761191424e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8962962165334016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.222370338482422e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.090471635124224e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738961367236608.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7484033703804016, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4488229751586914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.407288718249165e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9513278076029.912, + "step": 270, + "teaching_rms": 1174966171479.231, + "traffic_rms": 9513278076029.912 + }, + { + "batch_loss": 2.7780213356018066, + "innovation_rms": 1139758831902.594, + "instruction_rms": 0.005437255233057363, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4022301312079626e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1784928432876046e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7857657700417536.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.754822313785553, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5851456522941589, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.485865144929485e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8066520357666816e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9489466100023296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.250184546691318e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7814244716118016e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7857657700417536.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7482256293296814, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5848312377929688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.485865144929485e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9228215876941.95, + "step": 271, + "teaching_rms": 1139758831902.594, + "traffic_rms": 9228215876941.95 + }, + { + "batch_loss": 2.6164398193359375, + "innovation_rms": 1102483106465.7756, + "instruction_rms": 0.005715452036620022, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.457221243198505e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1926885338318474e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7064405695004672.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7552866339683533, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4481150805950165, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.556509626007552e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.707484247379149e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9916786354946048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.275217334079324e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.503281960026112e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7064405695004672.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7469819784164429, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44780072569847107, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.556509626007552e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8926408683274.986, + "step": 272, + "teaching_rms": 1102483106465.7756, + "traffic_rms": 8926408683274.986 + }, + { + "batch_loss": 2.8040590286254883, + "innovation_rms": 1063528933859.9899, + "instruction_rms": 0.0067898319856261515, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5126285203401605e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1999021651832682e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6350408721104896.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7614468932151794, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4490288496017456, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.62001329445929e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.60495672082432e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.025274728677376e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.297747014527222e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2529536566493184e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6350408721104896.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.746012270450592, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4493202567100525, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.62001329445929e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8611010098560.107, + "step": 273, + "teaching_rms": 1063528933859.9899, + "traffic_rms": 8611010098560.107 + }, + { + "batch_loss": 2.437579870223999, + "innovation_rms": 1023254869234.2441, + "instruction_rms": 0.007407017811423876, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5637889338195313e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2000338510549472e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5707747932766208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7653301358222961, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3764767050743103, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.677090832842752e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.499908263215104e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0504836097245184e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3180237269303296e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.027658140660531e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5707747932766208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7468864917755127, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3767680823802948, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.677090832842752e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8284927296095.702, + "step": 274, + "teaching_rms": 1023254869234.2441, + "traffic_rms": 8284927296095.702 + }, + { + "batch_loss": 2.874364137649536, + "innovation_rms": 981990810827.7897, + "instruction_rms": 0.02347550970496394, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.607409446311428e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.193204522714472e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5129295831760896.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7656329274177551, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3428504168987274, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.728383909271962e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.393106170458931e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0680204980649984e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.336272527574958e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8248923051196416e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5129295831760896.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7484674453735352, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3428691029548645, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.728383909271962e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7950826253289.283, + "step": 275, + "teaching_rms": 981990810827.7897, + "traffic_rms": 7950826253289.283 + }, + { + "batch_loss": 2.8250606060028076, + "innovation_rms": 940037884365.455, + "instruction_rms": 0.015289506159827585, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6623942361101107e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1797146127351268e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4608637884432384.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.760678768157959, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34431272745132446, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.77447062634496e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2852494484832256e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0785665755119616e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.352696482514862e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.642403010183168e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4608637884432384.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7500870823860168, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3445328176021576, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.77447062634496e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7611147403201.906, + "step": 276, + "teaching_rms": 940037884365.455, + "traffic_rms": 7611147403201.906 + }, + { + "batch_loss": 2.4455738067626953, + "innovation_rms": 897670557473.7886, + "instruction_rms": 0.012006625227099528, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7361138794123756e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1599986380707006e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4139999507251200.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.75323486328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49967917799949646, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.815870675104563e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1769724639576064e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0827690936369152e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.367478041960776e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4781627091648512e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4139999507251200.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7515961527824402, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4998992085456848, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.815870675104563e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7268114960145.778, + "step": 277, + "teaching_rms": 897670557473.7886, + "traffic_rms": 7268114960145.778 + }, + { + "batch_loss": 2.552199125289917, + "innovation_rms": 855138156680.5433, + "instruction_rms": 0.011951478091954681, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.806205821537747e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.134591079333155e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3718183789789184.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7505477666854858, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43036431074142456, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.853052206986035e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.06884816551936e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.08124191064064e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.380781445462098e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3303463630864384e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3718183789789184.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7520996928215027, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43035876750946045, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.853052206986035e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6923745836228.285, + "step": 278, + "teaching_rms": 855138156680.5433, + "traffic_rms": 6923745836228.285 + }, + { + "batch_loss": 2.8089118003845215, + "innovation_rms": 812665748870.2764, + "instruction_rms": 0.010054615729769893, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8733396650675405e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1041268450201385e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3338512170483712.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7538780570030212, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38150715827941895, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.886436987777843e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.961391519747277e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0745655987273728e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.392754439893811e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1973116838281216e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3338512170483712.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7509881854057312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3814123272895813, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.886436987777843e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6579862106161.649, + "step": 279, + "teaching_rms": 812665748870.2764, + "traffic_rms": 6579862106161.649 + }, + { + "batch_loss": 2.567772626876831, + "innovation_rms": 770455669127.511, + "instruction_rms": 0.021387628215082277, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.913901748527825e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0692708636752521e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2996774507642880.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7596375942230225, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5098440647125244, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.916404692588954e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855062732387123e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0632876588531712e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4035303410407834e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0775804510208e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2996774507642880.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7491747140884399, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5098280906677246, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.916404692588954e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6238102012316.371, + "step": 280, + "teaching_rms": 770455669127.511, + "traffic_rms": 6238102012316.371 + }, + { + "batch_loss": 2.532972812652588, + "innovation_rms": 728688562975.0425, + "instruction_rms": 0.014617583432167715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.952533501886595e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0307066329426864e+33, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2689180492627968.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.764049768447876, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39120879769325256, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.943296341822669e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7502703622029312e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0479235944677376e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.413228720792535e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9698224059187200.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2689180492627968.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7489060163497925, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39098888635635376, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.943296341822669e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5899928817985.97, + "step": 281, + "teaching_rms": 728688562975.0425, + "traffic_rms": 5899928817985.97 + }, + { + "batch_loss": 2.594910144805908, + "innovation_rms": 687525016799.1489, + "instruction_rms": 0.02137518312865902, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9958685536721306e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.891174957241269e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2412319116099584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7650308609008789, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46608710289001465, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.967419455137382e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6473746495766528e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0289573410111488e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.421957125130158e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728401653268480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2412319116099584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7530350685119629, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46586722135543823, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.967419455137382e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5566642433813.524, + "step": 282, + "teaching_rms": 687525016799.1489, + "traffic_rms": 5566642433813.524 + }, + { + "batch_loss": 2.467066764831543, + "innovation_rms": 647106409101.0566, + "instruction_rms": 0.03538317064262344, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0326520279846093e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.451700053594762e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2163119744876544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7619132995605469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3064148426055908, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.989050628426957e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5466904156110848e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0068421249073152e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.429812792113234e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7855561219506176.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2163119744876544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7559893131256104, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30640268325805664, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.989050628426957e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5239388031418.187, + "step": 283, + "teaching_rms": 647106409101.0566, + "traffic_rms": 5239388031418.187 + }, + { + "batch_loss": 2.4869887828826904, + "innovation_rms": 607556592646.6499, + "instruction_rms": 0.14912164545015086, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.061373470725374e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.995006951426845e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1938818701721600.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7564358115196228, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21899709105491638, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.00843896979456e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4484902833553408e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9820012151832576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4368826518798336e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7070004721745920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1938818701721600.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7577599287033081, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21919791400432587, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.00843896979456e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4919167072019.987, + "step": 284, + "teaching_rms": 607556592646.6499, + "traffic_rms": 4919167072019.987 + }, + { + "batch_loss": 2.623499870300293, + "innovation_rms": 568982566100.187, + "instruction_rms": 0.03059321527627628, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.080963194519552e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.527060974301824e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1736928261046272.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7535920143127441, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3767578601837158, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.025808676533043e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3530073621594112e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9548291045851136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4432457318282035e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6363004249571328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1736928261046272.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7552680969238281, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37692850828170776, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.025808676533043e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4606846731022.4375, + "step": 285, + "teaching_rms": 568982566100.187, + "traffic_rms": 4606846731022.4375 + }, + { + "batch_loss": 2.838550329208374, + "innovation_rms": 531475842029.5588, + "instruction_rms": 0.11800157119574428, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.112382838794879e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.053353933718979e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1555209536929792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7511909604072571, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32753926515579224, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.04136075311186e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.260438146777088e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9256921538232320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.448972469421998e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5726703770927104.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1555209536929792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7460969686508179, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3274790942668915, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.04136075311186e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4303168269304.314, + "step": 286, + "teaching_rms": 531475842029.5588, + "traffic_rms": 4303168269304.314 + }, + { + "batch_loss": 2.5661211013793945, + "innovation_rms": 495113627116.5103, + "instruction_rms": 0.28565724635526757, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1367695943820902e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.57887318061158e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1391647115968512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7488950490951538, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41555753350257874, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.055277306144358e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1709452017205248e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8949296653139968.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.454126430177198e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5154033232773120.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1391647115968512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.733802855014801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.415497362613678, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.055277306144358e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4008757031546.1, + "step": 287, + "teaching_rms": 495113627116.5103, + "traffic_rms": 4008757031546.1 + }, + { + "batch_loss": 2.584005117416382, + "innovation_rms": 459959836343.5791, + "instruction_rms": 0.46266752050715865, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.155400956353708e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.108051314032794e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1244427179786240.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.749926745891571, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.481788694858551, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.067721544387789e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0846597382406144e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8628550106087424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.458764994856878e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4638629641060352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1244427179786240.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.718867838382721, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48180556297302246, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.067721544387789e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3724129377184.266, + "step": 288, + "teaching_rms": 459959836343.5791, + "traffic_rms": 3724129377184.266 + }, + { + "batch_loss": 2.4493520259857178, + "innovation_rms": 426065960668.5452, + "instruction_rms": 0.32929382691828557, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.169508515171795e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.644756896605388e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111916768395264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7529140710830688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5564342737197876, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.078840355723674e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.001684083933184e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8297567578226688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.46293970306859e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4174766462205952.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111916768395264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7184092998504639, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5563388466835022, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.078840355723674e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3449702908414.7837, + "step": 289, + "teaching_rms": 426065960668.5452, + "traffic_rms": 3449702908414.7837 + }, + { + "batch_loss": 2.659411668777466, + "innovation_rms": 393472254593.99805, + "instruction_rms": 0.23012329559766917, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.179859867391492e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.192302965359499e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 992646264782848.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.757530927658081, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.481672465801239, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.088766884138189e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9220941523451904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7958995306283008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.466696940459131e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3757289735454720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 992646264782848.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7326428890228271, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48165562748908997, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.088766884138189e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3185803265597.3003, + "step": 290, + "teaching_rms": 393472254593.99805, + "traffic_rms": 3185803265597.3003 + }, + { + "batch_loss": 2.522209644317627, + "innovation_rms": 362208449242.5926, + "instruction_rms": 0.4961715545364079, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.18727483637026e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.7534628928393885e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 885292819021824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7531240582466125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4069271683692932, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.097619670728704e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8459416441454592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7615252430585856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4700786259093094e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3381560627691520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 885292819021824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7445454001426697, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40691035985946655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.097619670728704e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2932671723710.3066, + "step": 291, + "teaching_rms": 362208449242.5926, + "traffic_rms": 2932671723710.3066 + }, + { + "batch_loss": 2.37760066986084, + "innovation_rms": 332294980866.75653, + "instruction_rms": 0.24476881396062242, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.192067470116782e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.3305005618919e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 788665919864832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7459095120429993, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31852471828460693, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.10550608967762e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7732562482954240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7268541732487168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.473121867936563e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3043404430704640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 788665919864832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7481825947761536, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31841713190078735, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.10550608967762e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2690472733966.544, + "step": 292, + "teaching_rms": 332294980866.75653, + "traffic_rms": 2690472733966.544 + }, + { + "batch_loss": 2.5814783573150635, + "innovation_rms": 303743363366.4911, + "instruction_rms": 0.11992823972008551, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1976873489242522e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.925203635305014e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701693838753792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7446881532669067, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23116789758205414, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.112523207245824e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7040476284715008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6920860371779584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.47586102627926e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2739063853416448.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701693838753792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7370426654815674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23119261860847473, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.112523207245824e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2459300641337.3667, + "step": 293, + "teaching_rms": 303743363366.4911, + "traffic_rms": 2459300641337.3667 + }, + { + "batch_loss": 2.62662410736084, + "innovation_rms": 276557472592.5484, + "instruction_rms": 0.140716572977433, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2033157489468375e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.538921080865289e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 623411885768704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7466797828674316, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2959233224391937, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.118756922779238e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6383075168616448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6574010087243776.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.478326337507164e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2465157414387712.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 623411885768704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7210994362831116, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29595592617988586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.118756922779238e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2239186229074.108, + "step": 294, + "teaching_rms": 276557472592.5484, + "traffic_rms": 2239186229074.108 + }, + { + "batch_loss": 2.3634839057922363, + "innovation_rms": 250734035325.6073, + "instruction_rms": 0.10841803672247768, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2098375021669908e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.17260533869045e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 552951907090432.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7478203177452087, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2627227306365967, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.124286263676109e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5760114321522688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6229609007808512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.480544945813586e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2218641592418304.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 552951907090432.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7187484502792358, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2628174126148224, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.124286263676109e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2030103304968.8313, + "step": 295, + "teaching_rms": 250734035325.6073, + "traffic_rms": 2030103304968.8313 + }, + { + "batch_loss": 2.5185723304748535, + "innovation_rms": 226263474410.48053, + "instruction_rms": 0.2783746157483933, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.21985721419196e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.8268529411369284e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 489532419997696.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7498197555541992, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3811732232570648, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.12918166740009e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5171203975151616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5889100779356160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4825415902101504e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1996777406332928.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 489532419997696.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7346099019050598, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38118916749954224, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.12918166740009e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1831973917352.2976, + "step": 296, + "teaching_rms": 226263474410.48053, + "traffic_rms": 1831973917352.2976 + }, + { + "batch_loss": 2.3678412437438965, + "innovation_rms": 203130450169.26556, + "instruction_rms": 0.16246433594404752, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2274985451270963e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.501940103575989e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 432449989378048.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7533034086227417, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3306979537010193, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.133505840473702e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4615827659685888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5553765302140928.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.484338604526797e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1797099679121408.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 432449989378048.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7477221488952637, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.330679327249527, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.133505840473702e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1644674117975.5964, + "step": 297, + "teaching_rms": 203130450169.26556, + "traffic_rms": 1644674117975.5964 + }, + { + "batch_loss": 2.4705538749694824, + "innovation_rms": 181314675557.44254, + "instruction_rms": 0.0948200244237861, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.245978449371988e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.1978749499891405e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 381071476654080.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7548023462295532, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19231829047203064, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.137316335458714e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4093354820239360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5224729468207104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4859559174117786e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1617389657522176.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 381071476654080.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7468759417533875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1923573762178421, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.137316335458714e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1468039620862.6816, + "step": 298, + "teaching_rms": 181314675557.44254, + "traffic_rms": 1468039620862.6816 + }, + { + "batch_loss": 2.6305339336395264, + "innovation_rms": 160791371942.3479, + "instruction_rms": 0.2008260578597921, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2678741239275192e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.914430589178557e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 334826993352704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7497052550315857, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38380488753318787, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.140664691962675e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3603056922984448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4902977898807296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.487411739526431e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1455650718613504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 334826993352704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7320961952209473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3838234841823578, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.140664691962675e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1301869882368.801, + "step": 299, + "teaching_rms": 160791371942.3479, + "traffic_rms": 1301869882368.801 + }, + { + "batch_loss": 2.386336088180542, + "innovation_rms": 141531941859.79214, + "instruction_rms": 0.4573177578382843, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2734215724065096e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.651188442668455e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 293203626426368.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7430608868598938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4796379506587982, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.143596436638925e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3144120876924928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4589359386853376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4887218763504026e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1310085552799744.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 293203626426368.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7086266279220581, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4796220064163208, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.143596436638925e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1145933225217.243, + "step": 300, + "teaching_rms": 141531941859.79214, + "traffic_rms": 1145933225217.243 + }, + { + "batch_loss": 2.6240406036376953, + "innovation_rms": 123504442489.44261, + "instruction_rms": 0.26551561507529287, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.294119329043579e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.4075732551968292e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 255739666890752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7423151731491089, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2158840298652649, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.146153660166963e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2715661113491456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4284598171205632.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.489901102571192e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1179076970676224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 255739666890752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.719230055809021, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21595832705497742, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.146153660166963e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 999971026819.9326, + "step": 301, + "teaching_rms": 123504442489.44261, + "traffic_rms": 999971026819.9326 + }, + { + "batch_loss": 2.615520477294922, + "innovation_rms": 106674080876.0302, + "instruction_rms": 0.28256786204480083, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3117350171490386e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.18287649951932e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 222019526328320.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7485514283180237, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3968232274055481, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.148374158258995e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2316730860830720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3989303063478272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.490962131291996e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1061169246765056.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 222019526328320.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7334929704666138, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3968287408351898, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.148374158258995e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 863701664773.9491, + "step": 302, + "teaching_rms": 106674080876.0302, + "traffic_rms": 863701664773.9491 + }, + { + "batch_loss": 2.4371719360351562, + "innovation_rms": 91003677699.15459, + "instruction_rms": 1.6132167200682692, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.311935403143201e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.97628809862272e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191669190262784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7535519003868103, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3631998300552368, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15029057266647e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1946333283876864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3703975501103104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4919173320186266e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 955052281823232.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191669190262784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7391068339347839, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36298036575317383, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15029057266647e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 736823984135.872, + "step": 303, + "teaching_rms": 91003677699.15459, + "traffic_rms": 736823984135.872 + }, + { + "batch_loss": 2.6060378551483154, + "innovation_rms": 76454061519.53996, + "instruction_rms": 0.7945145092492792, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.30701179007402e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.786923699091465e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164351973523456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7435470819473267, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29994380474090576, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15193382715392e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1603431416463360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3429018137264128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.492777012672594e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 859547040219136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164351973523456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7257888913154602, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29972437024116516, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15193382715392e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 619021038801.6155, + "step": 304, + "teaching_rms": 76454061519.53996, + "traffic_rms": 619021038801.6155 + }, + { + "batch_loss": 2.688016653060913, + "innovation_rms": 62984507686.377884, + "instruction_rms": 0.7451005728040311, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.295873737284649e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.6138444974910743e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139764846034944.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7349134683609009, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30871596932411194, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.153331409512038e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1286957153910784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3164742088654848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4935504503832576e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 773592329486336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139764846034944.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7054195404052734, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3086060881614685, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.153331409512038e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 509962888673.41644, + "step": 305, + "teaching_rms": 62984507686.377884, + "traffic_rms": 509962888673.41644 + }, + { + "batch_loss": 2.4197282791137695, + "innovation_rms": 50553059861.03061, + "instruction_rms": 0.5964520231461684, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2813904203677696e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.4560785174625784e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117635027042304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7317370176315308, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.269302099943161, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.154507371557683e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 995819708743680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2911374720106496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.494246578682593e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 696233056272384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117635027042304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6938129663467407, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26939743757247925, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.154507371557683e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 409309916709.1968, + "step": 306, + "teaching_rms": 50553059861.03061, + "traffic_rms": 409309916709.1968 + }, + { + "batch_loss": 2.784821033477783, + "innovation_rms": 39116867241.48341, + "instruction_rms": 0.05475348429019719, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2828256953588777e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.3126384051105814e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97717015543808.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7341641187667847, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4679993987083435, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.155484906114253e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 728912925556736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2669067966087168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4948733003104256e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 626609757356032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97717015543808.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6950281262397766, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46810925006866455, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.155484906114253e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 316715192490.1475, + "step": 307, + "teaching_rms": 39116867241.48341, + "traffic_rms": 316715192490.1475 + }, + { + "batch_loss": 2.4955825805664062, + "innovation_rms": 28632481974.955135, + "instruction_rms": 0.07748647803770227, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2803417611527782e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1825363229693598e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79789822050304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7402223348617554, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4870750308036804, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15628291103785e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 485122528772096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2437903967846400.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4954371436170445e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 563948734644224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79789822050304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7088872194290161, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48718491196632385, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15628291103785e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 231826900155.84653, + "step": 308, + "teaching_rms": 28632481974.955135, + "traffic_rms": 231826900155.84653 + }, + { + "batch_loss": 2.6780924797058105, + "innovation_rms": 19056133598.31992, + "instruction_rms": 0.07472096777036707, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.275979723647484e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0647972965439108e+32, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63654556860416.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7465993762016296, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3412091135978699, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.156919425191117e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 263332313432064.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2217902052737024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.49594463695274e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 507553834336256.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63654556860416.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7163792252540588, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.341113805770874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.156919425191117e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 154290650019.80136, + "step": 309, + "teaching_rms": 19056133598.31992, + "traffic_rms": 154290650019.80136 + }, + { + "batch_loss": 2.7340192794799805, + "innovation_rms": 10343976696.304811, + "instruction_rms": 0.09453227234098706, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2687479607936942e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.58469175858706e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49132181913600.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7361464500427246, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3855075240135193, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15741076944978e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 62429857841152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2009024505577472.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4964012778756506e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 456798427414528.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49132181913600.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7132425308227539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3856239318847656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15741076944978e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 83751450789.6603, + "step": 310, + "teaching_rms": 10343976696.304811, + "traffic_rms": 83751450789.6603 + }, + { + "batch_loss": 2.8255417346954346, + "innovation_rms": 2452311836.951754, + "instruction_rms": 0.22474412331220844, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.262092616910766e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.626307594391989e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40723822936064.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.731191098690033, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3902926445007324, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.157771546702643e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 118688476823552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1811183279538176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4968125639439155e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 411118564540416.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40723822936064.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7048598527908325, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39018285274505615, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.157771546702643e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19855486358.77378, + "step": 311, + "teaching_rms": 2452311836.951754, + "traffic_rms": 19855486358.77378 + }, + { + "batch_loss": 2.4956440925598145, + "innovation_rms": 0.24824939755238862, + "instruction_rms": 0.24578650468353563, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.254280449356464e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.763676641524659e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33302708748288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7293559312820435, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.65464186668396, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.158014641851597e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 281694959566848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1630064911319040.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.497182618326139e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 370006701375488.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33302708748288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7002999186515808, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6545320749282837, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.158014641851597e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30662519413358164, + "step": 312, + "teaching_rms": 0.24824939755238862, + "traffic_rms": 0.183325255697421 + }, + { + "batch_loss": 2.5421769618988037, + "innovation_rms": 0.22849558651211638, + "instruction_rms": 0.22582474039980907, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2472644656596255e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.987308832301095e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26623374000128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7284650206565857, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46134692430496216, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.158152080805069e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 428400808296448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1467058420187136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.497515564190925e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 333006027882496.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26623374000128.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.706011950969696, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4612371325492859, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.158152080805069e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2847193830875436, + "step": 313, + "teaching_rms": 0.22849558651211638, + "traffic_rms": 0.17340289914809262 + }, + { + "batch_loss": 2.857010841369629, + "innovation_rms": 0.2629733804761913, + "instruction_rms": 0.2607647647585205, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2419163036631695e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.288577755642854e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20611705339904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7352421879768372, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32722046971321106, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15819417148457e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 560436055375872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1320352537903104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.497815181109494e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 299705401606144.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20611705339904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7146577835083008, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3272331953048706, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15819417148457e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3083140394696722, + "step": 314, + "teaching_rms": 0.2629733804761913, + "traffic_rms": 0.164497695688819 + }, + { + "batch_loss": 2.4180572032928467, + "innovation_rms": 0.19538486394825788, + "instruction_rms": 0.19256484228499007, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2397862747622605e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.6597201735067e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16932570923008.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7451134920120239, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3397540748119354, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15815036281815e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 679267801235456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1188317190160384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4980849050556826e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 269734851379200.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16932570923008.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7159736156463623, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3396743834018707, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15815036281815e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24686112642379734, + "step": 315, + "teaching_rms": 0.19538486394825788, + "traffic_rms": 0.15446401647809138 + }, + { + "batch_loss": 2.519407033920288, + "innovation_rms": 0.1167718826271365, + "instruction_rms": 0.11247073963171028, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2418537689393398e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.093748011084931e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20505344081920.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7479599714279175, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2372075915336609, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.158029244740403e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 786216379219968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1069485444300800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.498327828405944e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242761366241280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20505344081920.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7158330678939819, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23712792992591858, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.158029244740403e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.18395104964928485, + "step": 316, + "teaching_rms": 0.1167718826271365, + "traffic_rms": 0.14556208925080907 + }, + { + "batch_loss": 2.8352608680725098, + "innovation_rms": 0.07922098166780822, + "instruction_rms": 0.0735244365836975, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2481883303048643e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.584373403404569e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25996445089792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7399304509162903, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2997116148471832, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15783854819246e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 882470085984256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 962536866316288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.498546356341965e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 218485221228544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25996445089792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.71615070104599, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2996930480003357, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15783854819246e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.1562970438735198, + "step": 317, + "teaching_rms": 0.07922098166780822, + "traffic_rms": 0.13792423948975205 + }, + { + "batch_loss": 2.6539347171783447, + "innovation_rms": 0.060699453982810056, + "instruction_rms": 0.05391658165434553, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2582120280594842e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.125936159778178e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30938176159744.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7340241074562073, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.311968058347702, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.157585145121997e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 969098435493888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 866283159552000.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.49874289404543e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 196636689039360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30938176159744.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7252448201179504, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3119625449180603, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.157585145121997e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.1413581602464773, + "step": 318, + "teaching_rms": 0.060699453982810056, + "traffic_rms": 0.13067254354267863 + }, + { + "batch_loss": 2.7775936126708984, + "innovation_rms": 0.037115697172223155, + "instruction_rms": 0.025006498884555958, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2703289210754826e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.713342592157393e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35385421856768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7313686609268188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5248440504074097, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.157275907476685e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1047063903076352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 779654810042368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.498919846698025e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176973020135424.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35385421856768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7279397249221802, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5248385071754456, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.157275907476685e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.12569539709490624, + "step": 319, + "teaching_rms": 0.037115697172223155, + "traffic_rms": 0.12318456912879161 + }, + { + "batch_loss": 2.544624090194702, + "innovation_rms": 0.30536050445215623, + "instruction_rms": 0.30436367661171904, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.276313150548607e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.342008212049072e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39387588984832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7298579812049866, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5282554626464844, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15691598921728e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1117232864165888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701689342459904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4990792758840525e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 159275708055552.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39387588984832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7243906855583191, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5282499194145203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15691598921728e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.32536842402161475, + "step": 320, + "teaching_rms": 0.30536050445215623, + "traffic_rms": 0.11501101135206789 + }, + { + "batch_loss": 2.642643451690674, + "innovation_rms": 0.8504923976546954, + "instruction_rms": 0.8501835585657574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2827905109867889e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.0078074392011974e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42989145554944.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.733475923538208, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42437392473220825, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.156510544304538e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1180384922435584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 631520381370368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499222555993047e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 143348140605440.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42989145554944.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7114619612693787, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42437943816185, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.156510544304538e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8569948746353111, + "step": 321, + "teaching_rms": 0.8504923976546954, + "traffic_rms": 0.10783419421738684 + }, + { + "batch_loss": 2.7645628452301025, + "innovation_rms": 1.6724956363578671, + "instruction_rms": 1.672361538950998, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2956862705521132e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.7070266952810776e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46230117810176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7395860552787781, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5081912279129028, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.156063867705754e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1237221701058560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568368323100672.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499351405011927e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 129013326544896.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46230117810176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7044464349746704, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5081967115402222, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.156063867705754e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.6754909262245863, + "step": 322, + "teaching_rms": 1.6724956363578671, + "traffic_rms": 0.10235614782357601 + }, + { + "batch_loss": 2.7067832946777344, + "innovation_rms": 10.4514890658653, + "instruction_rms": 10.45146937450438, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.297587051278631e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.43632402575297e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49146530627584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7471573948860168, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.288137823343277, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.155580254388224e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1288374895771648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 511531477368832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499467540927611e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116111991373824.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49146530627584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7139139771461487, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28801795840263367, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.155580254388224e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10.451873690305979, + "step": 323, + "teaching_rms": 10.4514890658653, + "traffic_rms": 0.09193344120324692 + }, + { + "batch_loss": 2.788153886795044, + "innovation_rms": 1.509355583380826, + "instruction_rms": 1.50923217125048, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2947105914214154e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.192691623177673e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51770810171392.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7341794967651367, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4188368022441864, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.155063140325786e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1334412784435200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 460378316210176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4995719945322496e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104500790558720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51770810171392.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7184644341468811, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4187169373035431, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.155063140325786e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.5114308655946378, + "step": 324, + "teaching_rms": 1.509355583380826, + "traffic_rms": 0.08149658660080189 + }, + { + "batch_loss": 2.5986244678497314, + "innovation_rms": 3.0048355509828397, + "instruction_rms": 3.003482829765928, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2921353977502106e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.9734224608599056e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54658307457024.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7186758518218994, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46305596828460693, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.154516820485734e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1375846870810624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 414340461101056.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499666140215378e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94050707308544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54658307457024.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7122148275375366, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46313291788101196, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.154516820485734e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.315777090834804, + "step": 325, + "teaching_rms": 3.0048355509828397, + "traffic_rms": 1.4048020669343078 + }, + { + "batch_loss": 2.7142345905303955, + "innovation_rms": 2.43134672696067, + "instruction_rms": 2.233182253144118, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2893638038144942e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.7760801905953987e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57346990538752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7104048728942871, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41727349162101746, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.153943012854989e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1413137521704960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 372906408280064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499750665171763e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84645634899968.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57346990538752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7000753283500671, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4171180725097656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.153943012854989e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15.530941373977933, + "step": 326, + "teaching_rms": 2.43134672696067, + "traffic_rms": 15.369549028018495 + }, + { + "batch_loss": 2.7371511459350586, + "innovation_rms": 1.3614425828266252, + "instruction_rms": 1.3613799564884628, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2816493553561108e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.598472183625117e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59766231531520.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7097354531288147, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41899242997169495, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.153345153407386e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1446699067244544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 335615757385728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.49982694379094e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76181068054528.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59766231531520.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6960527896881104, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4188370406627655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.153345153407386e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.3634675151161286, + "step": 327, + "teaching_rms": 1.3614425828266252, + "traffic_rms": 0.07542289244562768 + }, + { + "batch_loss": 2.8585925102233887, + "innovation_rms": 0.1621076331010962, + "instruction_rms": 0.1607676361339429, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2658809842242683e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.4386249894411217e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61942953672704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.715774416923523, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.250189870595932, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.152725819123302e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1476904498495488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 302054178291712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.499895663267676e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68562957893632.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61942953672704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6941416263580322, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2500474452972412, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.152725819123302e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.18050339117982866, + "step": 328, + "teaching_rms": 0.1621076331010962, + "traffic_rms": 0.08206906110292393 + }, + { + "batch_loss": 2.901949167251587, + "innovation_rms": 3.129412839581492, + "instruction_rms": 0.13353817534864923, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.251330734537048e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.294762454229235e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63901383262208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7274417877197266, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38589978218078613, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.152086727989658e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1504089359777792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271848747040768.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4999575107967386e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61706659168256.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63901383262208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6919114589691162, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3859359622001648, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.152086727989658e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52.591189173947065, + "step": 329, + "teaching_rms": 3.129412839581492, + "traffic_rms": 52.59101963597794 + }, + { + "batch_loss": 2.4181978702545166, + "innovation_rms": 4.2843654383378365, + "instruction_rms": 0.09122698284118667, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2307946061092618e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1652862571633442e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65663330680832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7356060743331909, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2743515372276306, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.151430456986829e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1528555775197184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 244663868981248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500013173572895e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55535990734848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65663330680832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6914220452308655, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2743876874446869, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.151430456986829e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 73.04169054039433, + "step": 330, + "teaching_rms": 4.2843654383378365, + "traffic_rms": 73.04163357096185 + }, + { + "batch_loss": 2.8150670528411865, + "innovation_rms": 3.8976678982261443, + "instruction_rms": 0.08494676665374976, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2101692798006723e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0487576556255262e+31, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67248425271296.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.718410849571228, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5062344074249268, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.150757865108275e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1550575535652864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220197470339072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500062995193528e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49982388305920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67248425271296.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.689678430557251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5061983466148376, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.150757865108275e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 67.77996583862634, + "step": 331, + "teaching_rms": 3.8976678982261443, + "traffic_rms": 67.77991260997891 + }, + { + "batch_loss": 2.5383899211883545, + "innovation_rms": 3.5855074342832514, + "instruction_rms": 3.001954974279741, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1828192029376512e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.438818900629736e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68674337636352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7066293954849243, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34144890308380127, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.150071529334374e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1570393320062976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198177709883392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5001080064507904e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44984149475328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68674337636352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6888017058372498, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34141284227371216, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.150071529334374e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33.12667937716622, + "step": 332, + "teaching_rms": 3.5855074342832514, + "traffic_rms": 32.9903797049828 + }, + { + "batch_loss": 2.7582178115844727, + "innovation_rms": 2.6229723721538063, + "instruction_rms": 1.8595210065119963, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.157211852004655e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.494936768781598e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69956972576768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7048045992851257, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3296217620372772, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.149372308658586e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1588229379719168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178359942250496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5001485509420646e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40485733269504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69956972576768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6856446862220764, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3295930325984955, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.149372308658586e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30.648623046536468, + "step": 333, + "teaching_rms": 2.6229723721538063, + "traffic_rms": 30.592160375106847 + }, + { + "batch_loss": 2.4661145210266113, + "innovation_rms": 2.313245205753488, + "instruction_rms": 1.1634424902494493, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.138115121736581e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.645443333688602e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71110641057792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7065638303756714, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3419390916824341, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.148661062074368e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1604281819987968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160523949703168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500184972264735e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36437160361984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71110641057792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6862815618515015, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3418295383453369, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.148661062074368e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33.820888020226384, + "step": 334, + "teaching_rms": 2.313245205753488, + "traffic_rms": 33.80087081852907 + }, + { + "batch_loss": 2.4993834495544434, + "innovation_rms": 2.976094250069672, + "instruction_rms": 2.712373358735606, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1254848942293647e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.880898698088287e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72148236369920.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7123894095420837, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26677411794662476, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14793950756864e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1618729016229888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144471542988800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500217614016184e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32793442648064.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72148236369920.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6850241422653198, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26681363582611084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14793950756864e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 21.025395568320665, + "step": 335, + "teaching_rms": 2.976094250069672, + "traffic_rms": 20.849707186281346 + }, + { + "batch_loss": 2.504992723464966, + "innovation_rms": 5.387250453952036, + "instruction_rms": 5.387215284453433, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.120641682947965e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.19280894917204e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73081351569408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7182101607322693, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3090057671070099, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14720850413486e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1631731492847616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130024388689920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500247163391181e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29514096705536.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73081351569408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6871190071105957, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30898934602737427, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14720850413486e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.394137760412268, + "step": 336, + "teaching_rms": 5.387250453952036, + "traffic_rms": 0.273191823128673 + }, + { + "batch_loss": 2.6536405086517334, + "innovation_rms": 12.210099604418401, + "instruction_rms": 12.210082406448137, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.118873668250501e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.573528235593709e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73920422084608.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7125781178474426, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22133705019950867, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14646891076649e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1643433668116480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117021945626624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500273620389724e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26562686615552.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73920422084608.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6817026138305664, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22132065892219543, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14646891076649e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12.213396234422481, + "step": 337, + "teaching_rms": 12.210099604418401, + "traffic_rms": 0.28449144011140703 + }, + { + "batch_loss": 2.3480217456817627, + "innovation_rms": 2.707014541609278, + "instruction_rms": 2.706927530140097, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.120882475994448e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.0161753515880474e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74674843156480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7089110612869263, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34755441546440125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.145722445450445e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1653965599014912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105319745191936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500297672206582e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23906417115136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74674843156480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6809788942337036, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3475869297981262, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.145722445450445e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.720772702531137, + "step": 338, + "teaching_rms": 2.707014541609278, + "traffic_rms": 0.2741304850434237 + }, + { + "batch_loss": 2.4781270027160645, + "innovation_rms": 1.6748517603210962, + "instruction_rms": 1.4453249795361631, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1188750426400358e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.5145578768755336e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75353078890496.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7051712274551392, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27706021070480347, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.144969108186726e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1663444323401728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94787772350464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5003193188417536e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21515774984192.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75353078890496.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6812065839767456, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27685174345970154, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.144969108186726e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12.917502801937463, + "step": 339, + "teaching_rms": 1.6748517603210962, + "traffic_rms": 12.836390238507628 + }, + { + "batch_loss": 2.499730348587036, + "innovation_rms": 1.6285935139639691, + "instruction_rms": 1.165509386621166, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.115864854681092e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.063102058964835e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75962737754112.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7034767866134644, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2620009183883667, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.144209757968794e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1671975202193408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85308989243392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.50033856029524e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19364197695488.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75962737754112.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6833367347717285, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2619066834449768, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.144209757968794e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16.301851019554718, + "step": 340, + "teaching_rms": 1.6285935139639691, + "traffic_rms": 16.260133290620356 + }, + { + "batch_loss": 2.519381046295166, + "innovation_rms": 1.0118684014618158, + "instruction_rms": 0.2271594681561742, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1132819644284928e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.656791762398915e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76510673240064.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6978745460510254, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37516850233078003, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.143444394796646e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1679652993105920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76778085285888.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5003560837618074e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17427777716224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76510673240064.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6840035319328308, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37528812885284424, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.143444394796646e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13.568877167944722, + "step": 341, + "teaching_rms": 1.0118684014618158, + "traffic_rms": 13.566975574954656 + }, + { + "batch_loss": 2.5134124755859375, + "innovation_rms": 0.9170292025499232, + "instruction_rms": 0.43493585493548514, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1111368172427018e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.291112555935878e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77003050975232.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6942140460014343, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33120855689048767, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.142674736657203e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1686563058614272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69100273401856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5003718892414566e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15684999315456.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77003050975232.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6788390874862671, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3310471773147583, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.142674736657203e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11.113744487142503, + "step": 342, + "teaching_rms": 0.9170292025499232, + "traffic_rms": 11.105230633200845 + }, + { + "batch_loss": 2.5462918281555176, + "innovation_rms": 0.12959145161148414, + "instruction_rms": 0.12611510038006535, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.116258479843836e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.962001239895999e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77445415829504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6934288144111633, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27460235357284546, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.141899924557005e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1692782037041152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62190245642240.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5003859767341875e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14116499488768.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77445415829504.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6675876379013062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2746078372001648, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.141899924557005e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.15584609284737536, + "step": 343, + "teaching_rms": 0.12959145161148414, + "traffic_rms": 0.09155969723842357 + }, + { + "batch_loss": 2.5394885540008545, + "innovation_rms": 0.13271994210359522, + "instruction_rms": 0.1282746924383434, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1193061886370775e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.665801115906399e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77842775801856.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.699682891368866, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26256895065307617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14112167648297e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1698379184734208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55971221078016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500398689837384e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12704848805888.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77842775801856.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6558224558830261, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2624328136444092, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.14112167648297e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.1642298917342644, + "step": 344, + "teaching_rms": 0.13271994210359522, + "traffic_rms": 0.10255352183151027 + }, + { + "batch_loss": 2.5829100608825684, + "innovation_rms": 0.11659176099698278, + "instruction_rms": 0.10977341018907402, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1193082502213796e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.3992210345389048e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78199618797568.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7021921277046204, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2735082805156708, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.140339992435098e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1703416644501504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50374098550784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500410028551045e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11434363715584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78199618797568.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6601697206497192, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2736634314060211, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.140339992435098e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.15949524341010496, + "step": 345, + "teaching_rms": 0.11659176099698278, + "traffic_rms": 0.11570929797446959 + }, + { + "batch_loss": 2.5611305236816406, + "innovation_rms": 0.11321259606571218, + "instruction_rms": 0.10359309645354663, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1164286292682342e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.1592989764197325e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78519996514304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6977822184562683, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27824074029922485, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.139554872413389e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1707950250917888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45336689115136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5004203364725555e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10290927239168.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78519996514304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6719551682472229, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2779286801815033, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.139554872413389e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.16730245426375934, + "step": 346, + "teaching_rms": 0.11321259606571218, + "traffic_rms": 0.13137248270330984 + }, + { + "batch_loss": 2.4213850498199463, + "innovation_rms": 0.13492071692489893, + "instruction_rms": 0.12400621053371484, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1117863537368105e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.9433691090009048e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78807549607936.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.689745306968689, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21546721458435059, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.138767175411302e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1712030604066816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40803019784192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500429613601915e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9261833781248.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78807549607936.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.678769588470459, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2155710905790329, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.138767175411302e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.19400618904630604, + "step": 347, + "teaching_rms": 0.13492071692489893, + "traffic_rms": 0.1492006822965207 + }, + { + "batch_loss": 2.374376058578491, + "innovation_rms": 1.726318106089314, + "instruction_rms": 1.725193855718615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.104305688938283e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.7490321678776688e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79065566412800.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6818639636039734, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27765947580337524, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.137976901428838e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1715702935322624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36722716966912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500437859939123e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8335650193408.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79065566412800.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6813168525695801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2775041162967682, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.137976901428838e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.733521728742533, + "step": 348, + "teaching_rms": 1.726318106089314, + "traffic_rms": 0.16971654984370774 + }, + { + "batch_loss": 2.4515485763549805, + "innovation_rms": 2.8982849451314094, + "instruction_rms": 2.897321171672489, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0961884069072732e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.574129011536193e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79296982941696.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6797797083854675, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27631935477256775, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.137184050465997e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1719008046874624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33050444431360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500445419081564e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7502085226496.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79296982941696.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6824604272842407, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27660858631134033, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.137184050465997e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.9037637533455216, + "step": 349, + "teaching_rms": 2.8982849451314094, + "traffic_rms": 0.19332352110408488 + }, + { + "batch_loss": 2.5599238872528076, + "innovation_rms": 2.8770058992222958, + "instruction_rms": 2.8755242180930614, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0876003964606218e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.416716140605719e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79504466771968.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.685085117816925, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1706804782152176, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.136388622522778e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1721982580162560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29745399988224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500452291029238e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6751876284416.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79504466771968.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6747909188270569, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1707199513912201, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.136388622522778e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.883777344255877, + "step": 350, + "teaching_rms": 2.8770058992222958, + "traffic_rms": 0.21801945981852383 + }, + { + "batch_loss": 2.3926072120666504, + "innovation_rms": 1.8739276358999966, + "instruction_rms": 1.870568018666121, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0847804240132833e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.27504454165672e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79690408656896.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6946731805801392, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2854377031326294, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.13559147659264e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1724659686965248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26770858311680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500458475782144e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6076688236544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79690408656896.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6720172762870789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28533387184143066, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.13559147659264e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.8834703964852222, + "step": 351, + "teaching_rms": 1.8739276358999966, + "traffic_rms": 0.22008219907403398 + }, + { + "batch_loss": 2.441957950592041, + "innovation_rms": 1.1500207135543616, + "instruction_rms": 0.3892594745314658, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0809700664672256e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1475401177141934e+30, + "nonfinite_indices": [ + 15 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79856956080128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7013519406318665, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20191572606563568, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.134792612675584e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1727069029400576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24093772480512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.500463973340283e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5469019045888.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79856956080128.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6792261004447937, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20181189477443695, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.134792612675584e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003887234255671501, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11362972110509872, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14.864433406447581, + "step": 352, + "teaching_rms": 1.1500207135543616, + "traffic_rms": 14.85933570615397 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} diff --git a/results/kp_innovation_stability_grid/margin_0p003.json b/results/kp_innovation_stability_grid/margin_0p003.json new file mode 100644 index 0000000..ac15d15 --- /dev/null +++ b/results/kp_innovation_stability_grid/margin_0p003.json @@ -0,0 +1,33151 @@ +{ + "first_any_nonfinite_groups": { + "bn_running_var": [ + 16 + ] + }, + "first_any_nonfinite_step": 301, + "first_nonfinite_by_group": { + "bn_running_var": { + "indices": [ + 16 + ], + "step": 301 + } + }, + "first_nonfinite_by_metric": {}, + "first_nonfinite_metrics": [], + "first_training_failure_groups": {}, + "first_training_failure_step": null, + "initial_parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4011031687259674, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3956821858882904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1162850856781006, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 0.003337553469464183, + "max_applied_stability_margin": 0.0033375518396496773, + "max_positive_residual_soma_slope": 0.0, + "min_residual_soma_slope": -0.003337553469464183, + "mse": 7.272036461235194e-06, + "observations": 64, + "residual_traffic_rms_ratio": 0.15995903060998887, + "stability_margin": 0.003 + }, + "first_mse": 7.272036461235194e-06, + "last_mse": 7.272036461235194e-06, + "post_warmup_traffic_residual_rms_ratio": 0.15995903060998887, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_mixed_traffic_nonfinite_diagnosis_v1", + "provenance": { + "git_commit": "0ae690e7b3144da2c5b3384cd7169e86529fd26f", + "git_tracked_dirty": false + }, + "rule": "innovation", + "scope": "training_only_no_validation_or_test_evaluation", + "split": { + "dataset": "cifar10", + "input_layout": "NCHW", + "input_shape": [ + 3, + 32, + 32 + ], + "loader_seed": 0, + "normalization_mean": [ + 0.49140000343322754, + 0.4821999967098236, + 0.4465000033378601 + ], + "normalization_std": [ + 0.24699999392032623, + 0.2434999942779541, + 0.26159998774528503 + ], + "split_from_training_only": true, + "split_seed": 2027, + "test_examples": 10000, + "train_examples": 45000, + "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5", + "validation_class_counts": { + "0": 500, + "1": 500, + "2": 500, + "3": 500, + "4": 500, + "5": 500, + "6": 500, + "7": 500, + "8": 500, + "9": 500 + }, + "validation_examples": 5000, + "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b" + }, + "stability_margin": 0.003, + "test_evaluations": 0, + "traffic_calibration": { + "instruction_rms": [ + 0.008471464738249779, + 0.006617916747927666, + 0.003988116513937712, + 0.0043865954503417015, + 0.002713605063036084, + 0.003158062230795622, + 0.002004207344725728, + 0.003757704282179475, + 0.0027085733599960804, + 0.0030200635083019733, + 0.002056022873148322, + 0.002389939734712243, + 0.0016842411132529378, + 0.0034010226372629404, + 0.002549725119024515, + 0.0030542907770723104, + 0.0020918569061905146, + 0.002605808200314641, + 0.0018916188273578882 + ], + "realized_traffic_instruction_rms_ratio": [ + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 3.999999523162842, + 4.0, + 4.000000476837158, + 3.999999761581421, + 4.0, + 4.0, + 3.999999761581421, + 4.0, + 4.000000476837158, + 4.000000476837158, + 4.0, + 4.0, + 4.0 + ], + "target_ratio": 4.0, + "traffic_gain": [ + 0.0443243607878685, + 0.035817068070173264, + 0.013778294436633587, + 0.022762222215533257, + 0.007214079145342112, + 0.017218582332134247, + 0.004616298712790012, + 0.02042512409389019, + 0.007327491883188486, + 0.01611751690506935, + 0.004703814629465342, + 0.01256294921040535, + 0.003324519144371152, + 0.01805141195654869, + 0.006063441745936871, + 0.016296451911330223, + 0.0044285994954407215, + 0.013874167576432228, + 0.0036007820162922144 + ], + "unscaled_traffic_rms": [ + 0.7644973993301392, + 0.7390796542167664, + 1.157796859741211, + 0.7708553671836853, + 1.504616141319275, + 0.7336404919624329, + 1.736635684967041, + 0.7358984351158142, + 1.478581428527832, + 0.7495108246803284, + 1.7483876943588257, + 0.7609485983848572, + 2.0264477729797363, + 0.7536302804946899, + 1.6820316314697266, + 0.7496824264526367, + 1.8894071578979492, + 0.7512690424919128, + 2.10134220123291 + ] + }, + "trajectory": [ + { + "batch_loss": 2.9134020805358887, + "innovation_rms": 0.005032587581149961, + "instruction_rms": 0.004215397904306522, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40285658836364746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9875131845474243, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.814837694168091, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4058851897716522, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6322960257530212, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11819249391555786, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1819249391555786, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.007530927658081, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8693108558654785, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.814858913421631, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3794170618057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.632293164730072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1177096366882324, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017318262558628915, + "step": 1, + "teaching_rms": 0.005032587581149961, + "traffic_rms": 0.01680093259933294 + }, + { + "batch_loss": 2.383181571960449, + "innovation_rms": 0.009944468177359623, + "instruction_rms": 0.009731096474187725, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6955170631408691, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.993224859237671, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4888882637023926, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4080387353897095, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4674604535102844, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23896048963069916, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4423701763153076, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0017153024673462, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.358752965927124, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4888834953308105, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3808126151561737, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4674549698829651, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1187843084335327, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.016668615108117832, + "step": 2, + "teaching_rms": 0.009944468177359623, + "traffic_rms": 0.013539687767423866 + }, + { + "batch_loss": 2.533369541168213, + "innovation_rms": 0.025687857543152576, + "instruction_rms": 0.025618338704629883, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9180524349212646, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.720404624938965, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.244967222213745, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4115122854709625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3183801472187042, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4106180965900421, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.046145439147949, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9962931275367737, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.129619598388672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2449605464935303, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3904096484184265, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31837230920791626, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.130269169807434, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.027394911406576326, + "step": 3, + "teaching_rms": 0.025687857543152576, + "traffic_rms": 0.00968280049857625 + }, + { + "batch_loss": 2.6459836959838867, + "innovation_rms": 0.1106321972152315, + "instruction_rms": 0.11061645484900672, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0729758739471436, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.718188762664795, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.021787166595459, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4124864339828491, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1635015904903412, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9752329587936401, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7716590762138367, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.2077956199646, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.988315999507904, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.492952346801758, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0217785835266113, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3930494785308838, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16349166631698608, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2539069652557373, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.11082192485751999, + "step": 4, + "teaching_rms": 0.1106321972152315, + "traffic_rms": 0.006702016893912953 + }, + { + "batch_loss": 2.6751513481140137, + "innovation_rms": 0.7429880349451337, + "instruction_rms": 0.7429848569871106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1745615005493164, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.818680763244629, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.787534713745117, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4101289212703705, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18702761828899384, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.250150442123413, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.94925856590271, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.376998901367188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4496920108795166, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.81790542602539, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.787456512451172, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3871249854564667, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18698816001415253, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4381338357925415, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7430013826722146, + "step": 5, + "teaching_rms": 0.7429880349451337, + "traffic_rms": 0.005618643776325004 + }, + { + "batch_loss": 2.683047294616699, + "innovation_rms": 0.08270723545345797, + "instruction_rms": 0.08267924381914223, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.950665831565857, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.09476375579834, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.508668899536133, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40692755579948425, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29002413153648376, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3476288318634033, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.346168518066406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.969100952148438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.116243362426758, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.66551399230957, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.508580207824707, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3920237720012665, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2899615466594696, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5596683025360107, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.08395363441473758, + "step": 6, + "teaching_rms": 0.08270723545345797, + "traffic_rms": 0.014575276203898108 + }, + { + "batch_loss": 2.7092790603637695, + "innovation_rms": 0.010724507775494012, + "instruction_rms": 0.010385726737930407, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0963454246521, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.8814926147461, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.357487678527832, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4044579267501831, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4839426875114441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3721067905426025, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.588037490844727, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.418691635131836, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.370516777038574, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.542736053466797, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.357389450073242, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40037307143211365, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4841077923774719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5690042972564697, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.025308910364666513, + "step": 7, + "teaching_rms": 0.010724507775494012, + "traffic_rms": 0.023088053632045007 + }, + { + "batch_loss": 2.4888246059417725, + "innovation_rms": 0.005352727587323237, + "instruction_rms": 0.004661166044975754, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.164149284362793, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 255.35482788085938, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.321355819702148, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4056166112422943, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5972133874893188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.296573638916016, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.74666690826416, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.83891487121582, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.163341522216797, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.92824363708496, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.321249008178711, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4037473797798157, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5972205400466919, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.477365970611572, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.025638521187578867, + "step": 8, + "teaching_rms": 0.005352727587323237, + "traffic_rms": 0.025213286355138007 + }, + { + "batch_loss": 2.556396007537842, + "innovation_rms": 0.0037166655900639466, + "instruction_rms": 0.002835970222294144, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.527647018432617, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 536.90576171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.387642860412598, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4078368842601776, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3040487766265869, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.128400802612305, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.849884033203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.221620559692383, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.51327133178711, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.075358390808105, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.387527465820312, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4061666429042816, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3042426109313965, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.294744968414307, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021799951879044654, + "step": 9, + "teaching_rms": 0.0037166655900639466, + "traffic_rms": 0.021616941786698086 + }, + { + "batch_loss": 2.666769504547119, + "innovation_rms": 0.0032631036359407235, + "instruction_rms": 0.002419598076165226, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.101994514465332, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 863.05419921875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.529796123504639, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4218451976776123, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8099203109741211, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.874650478363037, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.904777526855469, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.464523315429688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.447300910949707, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.334668159484863, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.5296735763549805, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40610271692276, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8099014759063721, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.027317523956299, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.016490816394568994, + "step": 10, + "teaching_rms": 0.0032631036359407235, + "traffic_rms": 0.016315742550184573 + }, + { + "batch_loss": 2.3774166107177734, + "innovation_rms": 0.003254795229790769, + "instruction_rms": 0.0024884142594770236, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.59959888458252, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1155.075439453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.761796951293945, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45294374227523804, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7180604338645935, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.544996738433838, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.891387939453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.33870506286621, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.050105094909668, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.484246253967285, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.761667728424072, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4054393768310547, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7180405259132385, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.69282341003418, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.013826597158824514, + "step": 11, + "teaching_rms": 0.003254795229790769, + "traffic_rms": 0.013608536210805893 + }, + { + "batch_loss": 2.4111337661743164, + "innovation_rms": 0.0031869042785004236, + "instruction_rms": 0.0025210489253120524, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.021065711975098, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1339.161376953125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.077770709991455, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4718649983406067, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.544173538684845, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.147280216217041, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.775156021118164, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.837677001953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.424899101257324, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.1935453414917, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.077635288238525, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42968493700027466, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5441611409187317, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.300587177276611, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01165223683891572, + "step": 12, + "teaching_rms": 0.0031869042785004236, + "traffic_rms": 0.011374599667729949 + }, + { + "batch_loss": 2.450648546218872, + "innovation_rms": 0.003780906933157393, + "instruction_rms": 0.0033006992742589535, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.828142166137695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1401.019775390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.46356201171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47267282009124756, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44520533084869385, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6884613037109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.535444259643555, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.602876663208008, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.647623062133789, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.047343254089355, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.463449954986572, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45287495851516724, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44519245624542236, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.846377849578857, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009651415897167608, + "step": 13, + "teaching_rms": 0.003780906933157393, + "traffic_rms": 0.009075196499151889 + }, + { + "batch_loss": 2.4470107555389404, + "innovation_rms": 0.0035289725567818186, + "instruction_rms": 0.0029231062633938874, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.207108497619629, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1382.914794921875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.913687229156494, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47075238823890686, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.575528621673584, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.174945831298828, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.160446166992188, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.250011444091797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.775080680847168, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.156169414520264, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.913571357727051, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47289296984672546, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.575525164604187, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.336942672729492, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009280910750068113, + "step": 14, + "teaching_rms": 0.0035289725567818186, + "traffic_rms": 0.008808542472500226 + }, + { + "batch_loss": 2.486706495285034, + "innovation_rms": 0.005009498775737296, + "instruction_rms": 0.00454384550693962, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.5067720413208, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1307.7867431640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.554566383361816, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4720124900341034, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6068675518035889, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.612659454345703, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.64488410949707, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.844382286071777, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.849983215332031, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.620048522949219, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.554633617401123, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48807209730148315, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6068639755249023, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.77833080291748, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.009401125077112646, + "step": 15, + "teaching_rms": 0.005009498775737296, + "traffic_rms": 0.008228382187026281 + }, + { + "batch_loss": 2.389864683151245, + "innovation_rms": 0.010189759976338881, + "instruction_rms": 0.010007501429911015, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.904996871948242, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1207.4957275390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.133140563964844, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5007818937301636, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5417394042015076, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.006515502929688, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.984519958496094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.662263870239258, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.910775184631348, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.996420860290527, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.133209705352783, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5011630058288574, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5417357087135315, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.175492286682129, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01192692363443505, + "step": 16, + "teaching_rms": 0.010189759976338881, + "traffic_rms": 0.006489520017431994 + }, + { + "batch_loss": 2.4523613452911377, + "innovation_rms": 0.009482778145883587, + "instruction_rms": 0.009274878623483999, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.425341606140137, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111.7847900390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7196762561798096, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5249875783920288, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46890562772750854, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.360896110534668, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.1901912689209, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.54744815826416, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.965487480163574, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.196778774261475, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7197470664978027, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5134652853012085, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4689018428325653, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.532845497131348, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.011125065507331696, + "step": 17, + "teaching_rms": 0.009482778145883587, + "traffic_rms": 0.00615003864721607 + }, + { + "batch_loss": 2.6268060207366943, + "innovation_rms": 0.01434612529134289, + "instruction_rms": 0.014185628263882463, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.06570053100586, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1014.4560546875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3475210666656494, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5426435470581055, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3953171670436859, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.679744720458984, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.27529525756836, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.542301177978516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.014728546142578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.44973087310791, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3475935459136963, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5262220501899719, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3953900933265686, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.854368209838867, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.015440351673115032, + "step": 18, + "teaching_rms": 0.01434612529134289, + "traffic_rms": 0.006118550390429117 + }, + { + "batch_loss": 2.4115519523620605, + "innovation_rms": 0.01144809614229914, + "instruction_rms": 0.011264935800902173, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.753857612609863, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063.0247802734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0125482082366943, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5551297068595886, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6080416440963745, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.966611862182617, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.251888275146484, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.530037879943848, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.059045791625977, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.271183013916016, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0126218795776367, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.535926103591919, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.607783317565918, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.14363956451416, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.012712281415732034, + "step": 19, + "teaching_rms": 0.01144809614229914, + "traffic_rms": 0.005898093843307749 + }, + { + "batch_loss": 2.4810826778411865, + "innovation_rms": 0.004195158037537646, + "instruction_rms": 0.003694536228830557, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.476689338684082, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1185.2945556640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7110424041748047, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5622638463973999, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7733966112136841, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.224692344665527, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.130823135375977, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.605904579162598, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.098931312561035, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.66251277923584, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7111170291900635, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5386669635772705, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7732793092727661, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.40388298034668, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006836159200346915, + "step": 20, + "teaching_rms": 0.004195158037537646, + "traffic_rms": 0.005759391464533623 + }, + { + "batch_loss": 2.6561341285705566, + "innovation_rms": 0.00390517687661701, + "instruction_rms": 0.003363046179421683, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.251701354980469, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1314.4453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.43966007232666, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.565222978591919, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7746868133544922, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.456862449645996, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.921863555908203, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.697919845581055, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.134828567504883, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.017308235168457, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4397358894348145, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5438457727432251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7745678424835205, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.6379976272583, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006650494817575689, + "step": 21, + "teaching_rms": 0.00390517687661701, + "traffic_rms": 0.005746952117364737 + }, + { + "batch_loss": 2.47926926612854, + "innovation_rms": 0.0033035772169142897, + "instruction_rms": 0.0026776878386386136, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.89649772644043, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1434.8912353515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1953916549682617, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5722928643226624, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5818239450454712, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.665711402893066, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.633800506591797, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.854836940765381, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.167135238647461, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.315577030181885, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1954681873321533, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5493986010551453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5815547704696655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.848594665527344, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006495282038977906, + "step": 22, + "teaching_rms": 0.0033035772169142897, + "traffic_rms": 0.005920911677052447 + }, + { + "batch_loss": 2.735363483428955, + "innovation_rms": 0.0032266704958424917, + "instruction_rms": 0.0025560871729717423, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.325183868408203, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1553.015380859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9755281209945679, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5773526430130005, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5241467356681824, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.853568077087402, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.27454376220703, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.077362537384033, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.196211814880371, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1098103523254395, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9756053686141968, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5623118281364441, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5238746404647827, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.038022994995117, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.007377701353366132, + "step": 23, + "teaching_rms": 0.0032266704958424917, + "traffic_rms": 0.006930519050243815 + }, + { + "batch_loss": 2.553990602493286, + "innovation_rms": 0.008727213111804818, + "instruction_rms": 0.008506640753066669, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.129941940307617, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1666.7464599609375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.777631163597107, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6202973127365112, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6593894958496094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.023048400878906, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.851213455200195, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.369729518890381, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.222380638122559, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.397741794586182, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7777092456817627, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5774179100990295, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6597332954406738, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.208398818969727, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010022916862048438, + "step": 24, + "teaching_rms": 0.008727213111804818, + "traffic_rms": 0.005308798462991069 + }, + { + "batch_loss": 2.545802354812622, + "innovation_rms": 0.0053513383751626915, + "instruction_rms": 0.004984574680429985, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.479226112365723, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1795.797607421875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.599506139755249, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6824257373809814, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7300738096237183, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.176535606384277, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.370216369628906, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.732756614685059, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.245932579040527, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.757967472076416, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5995848178863525, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5920913219451904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7304205894470215, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.361624717712402, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006984398454455372, + "step": 25, + "teaching_rms": 0.0053513383751626915, + "traffic_rms": 0.004897199514744301 + }, + { + "batch_loss": 2.595956802368164, + "innovation_rms": 0.004965520183061101, + "instruction_rms": 0.004550657108521816, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.53748607635498, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1913.9364013671875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4391776323318481, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.722762942314148, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5471009016036987, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.31456184387207, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.837318420410156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.275420665740967, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.267128944396973, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.182170391082764, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4392569065093994, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5953822731971741, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5470007061958313, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.499414443969727, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006820707747921053, + "step": 26, + "teaching_rms": 0.004965520183061101, + "traffic_rms": 0.005088053747863802 + }, + { + "batch_loss": 2.580071210861206, + "innovation_rms": 0.003677458884481769, + "instruction_rms": 0.003093431621302441, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.559106826782227, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2003.5703125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3878536224365234, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7527413368225098, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8083950281143188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.43867301940918, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.25770950317383, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.209517002105713, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.286206245422363, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.976695537567139, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3880224227905273, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6172155737876892, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8081746101379395, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.623310089111328, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.005979670141626867, + "step": 27, + "teaching_rms": 0.003677458884481769, + "traffic_rms": 0.00512438602532275 + }, + { + "batch_loss": 2.62861967086792, + "innovation_rms": 0.004175261545819965, + "instruction_rms": 0.0036337112731510933, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.305243492126465, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2095.245361328125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6300779581069946, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7833973169326782, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7774322032928467, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.55025863647461, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.63606262207031, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.374323844909668, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.303375244140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.682098865509033, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6300151348114014, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6541928648948669, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7775590419769287, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.734700202941895, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0063388557045451705, + "step": 28, + "teaching_rms": 0.004175261545819965, + "traffic_rms": 0.0052042514186117494 + }, + { + "batch_loss": 2.4183197021484375, + "innovation_rms": 0.00471325181862454, + "instruction_rms": 0.004248675970188885, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.963194847106934, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2198.588623046875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8994343280792236, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.813633918762207, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7130439281463623, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.650569915771484, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.97658157348633, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.738422393798828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.318827629089355, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.269129753112793, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8993858098983765, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6884320974349976, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7128601670265198, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.834834098815918, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006701212373846554, + "step": 29, + "teaching_rms": 0.00471325181862454, + "traffic_rms": 0.00519390097019411 + }, + { + "batch_loss": 2.4105985164642334, + "innovation_rms": 0.004175176733556675, + "instruction_rms": 0.0035474943300073704, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.9818115234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2275.47314453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.190159320831299, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8379935026168823, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4350247383117676, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.74073314666748, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.28304672241211, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.214066505432129, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.332735061645508, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.478825569152832, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1904077529907227, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7201142907142639, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43505463004112244, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.924836158752441, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006525183513879746, + "step": 30, + "teaching_rms": 0.004175176733556675, + "traffic_rms": 0.005481093553232488 + }, + { + "batch_loss": 2.6399600505828857, + "innovation_rms": 0.003999589406219912, + "instruction_rms": 0.003156792587712283, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.092191696166992, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2363.721923828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.106741428375244, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8572355508804321, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.475011944770813, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.821763038635254, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.55886459350586, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.710826396942139, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.345251083374023, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.854059219360352, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.106692314147949, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7513348460197449, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47524237632751465, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.005718231201172, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006379198959272609, + "step": 31, + "teaching_rms": 0.003999589406219912, + "traffic_rms": 0.005543542777430112 + }, + { + "batch_loss": 2.4447336196899414, + "innovation_rms": 0.004918227153974521, + "instruction_rms": 0.004315210293283674, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.831573486328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2452.54150390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.146794319152832, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.874858021736145, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6424421668052673, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.894571304321289, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.80710220336914, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.171438217163086, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.356515884399414, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.076404571533203, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.146854400634766, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7794262766838074, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6426757574081421, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.078392028808594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.006852093191662753, + "step": 32, + "teaching_rms": 0.004918227153974521, + "traffic_rms": 0.005318871018738088 + }, + { + "batch_loss": 2.553338050842285, + "innovation_rms": 0.02616920418026258, + "instruction_rms": 0.026039330658840575, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.813968658447266, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2522.400634765625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4427080154418945, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8791693449020386, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.881649374961853, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.959980010986328, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.030513763427734, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.493828296661377, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.366654396057129, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.8394775390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4427080154418945, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.803958535194397, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8817770481109619, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.143677711486816, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02661582130762899, + "step": 33, + "teaching_rms": 0.02616920418026258, + "traffic_rms": 0.005514740707339414 + }, + { + "batch_loss": 2.6185712814331055, + "innovation_rms": 0.031050996179704648, + "instruction_rms": 0.030947408566878977, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.970197677612305, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2615.5654296875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.151602268218994, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8896123766899109, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4784935712814331, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.018728256225586, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.23158645629883, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.356487274169922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.375779151916504, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.73900604248047, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.151602268218994, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8187700510025024, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4786216616630554, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.202313423156738, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03144923191132248, + "step": 34, + "teaching_rms": 0.031050996179704648, + "traffic_rms": 0.0056014033133125465 + }, + { + "batch_loss": 2.544434070587158, + "innovation_rms": 0.014617836008768538, + "instruction_rms": 0.01436311545717594, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.694189071655273, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2683.395751953125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.420884132385254, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9065348505973816, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4133628308773041, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.071481704711914, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.41255187988281, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.778570175170898, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.389724731445312, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.220232009887695, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.420854568481445, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8333011269569397, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41359639167785645, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.254963874816895, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.015546733016270646, + "step": 35, + "teaching_rms": 0.014617836008768538, + "traffic_rms": 0.0059539606003954325 + }, + { + "batch_loss": 2.718416929244995, + "innovation_rms": 0.008368202311008719, + "instruction_rms": 0.007669086642150511, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.059831619262695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2750.409912109375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.560401916503906, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9299095273017883, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5931472182273865, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.118839263916016, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.57542037963867, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.981626510620117, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.887874603271484, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.981494903564453, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.560372352600098, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8356553316116333, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5932558178901672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.302227020263672, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010251201359474977, + "step": 36, + "teaching_rms": 0.008368202311008719, + "traffic_rms": 0.006813581664422419 + }, + { + "batch_loss": 2.492528200149536, + "innovation_rms": 0.007014625944474726, + "instruction_rms": 0.006064850905964013, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.429508209228516, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2789.216064453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.86216926574707, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9638930559158325, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7254465222358704, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.16133975982666, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.72200012207031, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.021899700164795, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.83858871459961, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.507144927978516, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.862170219421387, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8296882510185242, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7255553603172302, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.344640731811523, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.008917383006707061, + "step": 37, + "teaching_rms": 0.007014625944474726, + "traffic_rms": 0.006536073098953408 + }, + { + "batch_loss": 3.0039165019989014, + "innovation_rms": 0.00809082984944477, + "instruction_rms": 0.006609173621579656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.908029556274414, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2837.84326171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.631509780883789, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9599717259407043, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46405690908432007, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.199468612670898, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.85392379760742, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9968156814575195, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.762418746948242, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.238304138183594, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.631510734558105, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8473935127258301, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46402406692504883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.382689476013184, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.010029943424533465, + "step": 38, + "teaching_rms": 0.00809082984944477, + "traffic_rms": 0.007556575177304682 + }, + { + "batch_loss": 2.547844409942627, + "innovation_rms": 0.020643433635608225, + "instruction_rms": 0.019932509260651852, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.49517250061035, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2888.824951171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.095413208007812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9309350252151489, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.026196002960205, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.233662605285645, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.97265625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4121904373168945, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.760578155517578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.98160171508789, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.095414161682129, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.853959858417511, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.026163101196289, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.41680908203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021634273637308913, + "step": 39, + "teaching_rms": 0.020643433635608225, + "traffic_rms": 0.008413454233111862 + }, + { + "batch_loss": 2.845684766769409, + "innovation_rms": 0.028718062531508076, + "instruction_rms": 0.027989387478145444, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.39763069152832, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2929.493896484375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.669869422912598, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.916434109210968, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8384279608726501, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.264314651489258, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.07951354980469, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.3630852699279785, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.11180114746094, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.512229919433594, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.66988754272461, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8569803833961487, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8381552696228027, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.447392463684082, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.029535415850416497, + "step": 40, + "teaching_rms": 0.028718062531508076, + "traffic_rms": 0.009444651166420567 + }, + { + "batch_loss": 2.7636501789093018, + "innovation_rms": 0.008757881151688062, + "instruction_rms": 0.006151898563154535, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.589563369750977, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2953.781005859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.073614120483398, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9199950695037842, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7200359106063843, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.291778564453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.17568588256836, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.121078491210938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.7818717956543, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.70069885253906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.073583602905273, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8850176930427551, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7200313806533813, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.474793434143066, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.014651962461163523, + "step": 41, + "teaching_rms": 0.008757881151688062, + "traffic_rms": 0.013303055100489751 + }, + { + "batch_loss": 3.132093906402588, + "innovation_rms": 0.008763771286865655, + "instruction_rms": 0.005342682194028539, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.09446716308594, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3012.553955078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.565156936645508, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9422749280929565, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43133485317230225, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.316373825073242, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.26224136352539, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.1399507522583, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.852176666259766, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.70303344726562, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.56515884399414, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9266519546508789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43165266513824463, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.499329566955566, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.015272772715104892, + "step": 42, + "teaching_rms": 0.008763771286865655, + "traffic_rms": 0.014311905141427368 + }, + { + "batch_loss": 2.7953288555145264, + "innovation_rms": 0.013003319265499042, + "instruction_rms": 0.009711171322517797, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.80187225341797, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3067.1171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.382736206054688, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9605444669723511, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3666638433933258, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.338386535644531, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.34014129638672, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.163256645202637, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.86860275268555, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.16426086425781, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.38273811340332, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9286640286445618, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36684727668762207, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.521286964416504, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018005228004698712, + "step": 43, + "teaching_rms": 0.013003319265499042, + "traffic_rms": 0.015165793875102017 + }, + { + "batch_loss": 3.047285795211792, + "innovation_rms": 0.015248060271256409, + "instruction_rms": 0.009798885216410707, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.63763427734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3097.8984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.778675079345703, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9089886546134949, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3508516550064087, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.358074188232422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.41025161743164, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.300484657287598, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.35635375976562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.87750244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.778675079345703, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9135449528694153, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.350823163986206, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.540923118591309, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018461526970091856, + "step": 44, + "teaching_rms": 0.015248060271256409, + "traffic_rms": 0.01565445254678963 + }, + { + "batch_loss": 3.4712986946105957, + "innovation_rms": 0.0350536770321577, + "instruction_rms": 0.03369678200340746, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.5442008972168, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3120.3623046875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.009557723999023, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8964021801948547, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2081328630447388, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.450727462768555, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.473350524902344, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.67985725402832, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.94634246826172, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105.89990234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.0095157623291, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9009580612182617, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.208330512046814, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.450504302978516, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03997319398827862, + "step": 45, + "teaching_rms": 0.0350536770321577, + "traffic_rms": 0.02151618173668997 + }, + { + "batch_loss": 4.247214317321777, + "innovation_rms": 0.011450093084856174, + "instruction_rms": 0.006076391093252892, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.59587097167969, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3767.36279296875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.20289421081543, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9782421588897705, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8597445487976074, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.34429931640625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.5301399230957, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.064506530761719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 88.146728515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 102.00385284423828, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.20285415649414, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9827976226806641, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8599424362182617, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.34407615661621, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.025617565572914107, + "step": 46, + "teaching_rms": 0.011450093084856174, + "traffic_rms": 0.02490023012375797 + }, + { + "batch_loss": 4.390831470489502, + "innovation_rms": 0.016020220020732983, + "instruction_rms": 0.011749720086692065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.75164031982422, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6572.27978515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.075485229492188, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0271756649017334, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48933476209640503, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.871002197265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.58124923706055, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.57795238494873, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.10897064208984, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.62242889404297, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.0754451751709, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0317306518554688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48933032155036926, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.870779037475586, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.029971894319349292, + "step": 47, + "teaching_rms": 0.016020220020732983, + "traffic_rms": 0.027575469804105746 + }, + { + "batch_loss": 4.365326404571533, + "innovation_rms": 0.026489274499608324, + "instruction_rms": 0.023347044432245886, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.81744384765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9323.5595703125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.569843292236328, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0501924753189087, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49318927526474, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.04045295715332, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.62724685668945, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.126519203186035, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.184326171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.75357818603516, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.56984519958496, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0547469854354858, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49350282549858093, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.04022979736328, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03780134805166639, + "step": 48, + "teaching_rms": 0.026489274499608324, + "traffic_rms": 0.02973449043987266 + }, + { + "batch_loss": 3.654695749282837, + "innovation_rms": 0.02025591886206901, + "instruction_rms": 0.012342594427494427, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 173.17286682128906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11053.31640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.86771774291992, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0509788990020752, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6182851791381836, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.70368194580078, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.668643951416016, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.071314811706543, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.60890197753906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.2457275390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.867774963378906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0555330514907837, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6181297898292542, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.703458786010742, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.033998354434612, + "step": 49, + "teaching_rms": 0.02025591886206901, + "traffic_rms": 0.03168320152381434 + }, + { + "batch_loss": 3.135939598083496, + "innovation_rms": 0.03203510762112668, + "instruction_rms": 0.02318868879347348, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 185.4970703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11062.89453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.686859130859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.054850697517395, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7116785049438477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.480403900146484, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.705902099609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.264183044433594, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.0679473876953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.59039306640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.68682098388672, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0594043731689453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7119318246841431, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.480180740356445, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04062616923935462, + "step": 50, + "teaching_rms": 0.03203510762112668, + "traffic_rms": 0.03336024297286709 + }, + { + "batch_loss": 3.1825554370880127, + "innovation_rms": 0.026335784781854606, + "instruction_rms": 0.01932522605208414, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.15451049804688, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11015.5830078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.1868896484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0796067714691162, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8944852352142334, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.264347076416016, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.739437103271484, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.537685871124268, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.65248107910156, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 165.84535217285156, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.18695831298828, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0841600894927979, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8947386145591736, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.20984649658203, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.035038851045862685, + "step": 51, + "teaching_rms": 0.026335784781854606, + "traffic_rms": 0.029229298197341172 + }, + { + "batch_loss": 2.593303918838501, + "innovation_rms": 0.03235772925231196, + "instruction_rms": 0.008389012281147057, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.63140869140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10738.8583984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.77613067626953, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.103185772895813, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9540000557899475, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.853120803833008, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.769615173339844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.004308700561523, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 175.52871704101562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.7623748779297, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.77608489990234, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1077386140823364, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9542535543441772, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.85289764404297, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.041351080479929136, + "step": 52, + "teaching_rms": 0.03235772925231196, + "traffic_rms": 0.04049206065724255 + }, + { + "batch_loss": 2.3610973358154297, + "innovation_rms": 0.030370638932891093, + "instruction_rms": 0.009814777668682268, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 209.51454162597656, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14353.541015625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.35948944091797, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1214303970336914, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8795926570892334, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.08207130432129, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.796775817871094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.173280715942383, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 202.1959228515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.6719970703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.359466552734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1259827613830566, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8798462748527527, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.122419357299805, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0514848293531622, + "step": 53, + "teaching_rms": 0.030370638932891093, + "traffic_rms": 0.050541691262624484 + }, + { + "batch_loss": 3.0322768688201904, + "innovation_rms": 0.032739887412616596, + "instruction_rms": 0.00944852166278402, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 263.0397644042969, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22168.59375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.442787170410156, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.130142092704773, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7038282752037048, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.697998046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.82122039794922, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.137044906616211, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 231.2354736328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 290.39544677734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.44276428222656, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1346940994262695, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7036235332489014, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.643516540527344, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.05831377206481453, + "step": 54, + "teaching_rms": 0.032739887412616596, + "traffic_rms": 0.05754843517583339 + }, + { + "batch_loss": 2.339446783065796, + "innovation_rms": 0.03778258873897507, + "instruction_rms": 0.0034282084950123435, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 322.3948974609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 30872.703125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.12713623046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1364855766296387, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6796102523803711, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.31071090698242, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.84321975708008, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.0341215133667, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 264.1524353027344, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 329.16961669921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.12718963623047, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.141037106513977, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6794053912162781, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.2562370300293, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06512990184244395, + "step": 55, + "teaching_rms": 0.03778258873897507, + "traffic_rms": 0.0650398040349616 + }, + { + "batch_loss": 2.439615249633789, + "innovation_rms": 0.051558904542795755, + "instruction_rms": 0.007403607238409676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 376.95648193359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 36019.21875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.00940704345703, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1473190784454346, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46987974643707275, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.944664001464844, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.86302185058594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.991199493408203, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 306.5439758300781, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 423.9153137207031, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.00930786132812, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1518701314926147, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46967488527297974, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.890193939208984, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0736180653057624, + "step": 56, + "teaching_rms": 0.051558904542795755, + "traffic_rms": 0.07324420511832026 + }, + { + "batch_loss": 2.7178752422332764, + "innovation_rms": 0.05909676037910553, + "instruction_rms": 0.015153413590911997, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 401.0545959472656, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38416.5390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.94449615478516, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1616315841674805, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40635520219802856, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.540470123291016, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.88084411621094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.15848159790039, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 359.2872009277344, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 527.43212890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.944580078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166182279586792, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40637457370758057, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.48600387573242, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.07710771881407207, + "step": 57, + "teaching_rms": 0.05909676037910553, + "traffic_rms": 0.07560368319165729 + }, + { + "batch_loss": 2.6427817344665527, + "innovation_rms": 0.08083165945815908, + "instruction_rms": 0.0261889891640484, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 469.53363037109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 45240.75390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.75796508789062, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.175717830657959, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4581460654735565, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.310428619384766, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.896881103515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.7728271484375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 429.1938781738281, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 699.0668334960938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.75806427001953, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1802680492401123, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4581654369831085, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.33393859863281, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.10373912468528985, + "step": 58, + "teaching_rms": 0.08083165945815908, + "traffic_rms": 0.10037922764687267 + }, + { + "batch_loss": 2.6655642986297607, + "innovation_rms": 0.09066593570016225, + "instruction_rms": 0.023250037772072504, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 550.097900390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 55907.66796875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156.7481689453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1841485500335693, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49948132038116455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.08900451660156, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.91131591796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.36222839355469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 516.2156372070312, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 870.2175903320312, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156.74827575683594, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1886982917785645, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49950072169303894, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.087730407714844, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.12378176318859907, + "step": 59, + "teaching_rms": 0.09066593570016225, + "traffic_rms": 0.12157914504757174 + }, + { + "batch_loss": 2.8395540714263916, + "innovation_rms": 0.11448457102973987, + "instruction_rms": 0.05977616737698357, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 585.0537109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 62926.8984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.2338104248047, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1831183433532715, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6064733266830444, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.85709381103516, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.465702056884766, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.981689453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 619.9638671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1037.4820556640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.23374938964844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.187667727470398, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6064602136611938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.88059997558594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.142221618890736, + "step": 60, + "teaching_rms": 0.11448457102973987, + "traffic_rms": 0.12905092468485937 + }, + { + "batch_loss": 2.537301540374756, + "innovation_rms": 0.11368813010972298, + "instruction_rms": 0.029219442842263806, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.910400390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 102194.53125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.13821411132812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.175146222114563, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7259321808815002, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.11797332763672, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.18775939941406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.2205581665039, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 737.0541381835938, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1170.9029541015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.13821411132812, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1796951293945312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7259190678596497, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.08907318115234, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.18031512077817372, + "step": 61, + "teaching_rms": 0.11368813010972298, + "traffic_rms": 0.17793233146304502 + }, + { + "batch_loss": 2.640613317489624, + "innovation_rms": 0.14194659224143316, + "instruction_rms": 0.04476266441054064, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1002.3278198242188, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 149818.625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.614501953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1643964052200317, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.520012378692627, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.76869201660156, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.15672302246094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.68963623046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 874.3414916992188, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1372.87353515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.614501953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1689448356628418, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5199437141418457, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.76741790771484, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.197273150795289, + "step": 62, + "teaching_rms": 0.14194659224143316, + "traffic_rms": 0.1921279892945194 + }, + { + "batch_loss": 2.8962724208831787, + "innovation_rms": 0.19458345233438518, + "instruction_rms": 0.020553067624707027, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1124.315185546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 208728.265625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 254.76300048828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.175925612449646, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5247342586517334, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.71875762939453, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.92265319824219, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.65928649902344, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1058.423828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1840.8232421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 254.7630615234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1804735660552979, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5246873497962952, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.58049774169922, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24746602430336212, + "step": 63, + "teaching_rms": 0.19458345233438518, + "traffic_rms": 0.2466111701561511 + }, + { + "batch_loss": 2.5692477226257324, + "innovation_rms": 0.1730873242763517, + "instruction_rms": 0.04489582304734378, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193.8414306640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 294824.09375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 504.21295166015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.184571385383606, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42213016748428345, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.18853759765625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 88.73253631591797, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158.0988311767578, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1255.2955322265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1968.717041015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 504.2130432128906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1891188621520996, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.422328919172287, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.1180419921875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29892834655506717, + "step": 64, + "teaching_rms": 0.1730873242763517, + "traffic_rms": 0.2955377272368427 + }, + { + "batch_loss": 2.780534505844116, + "innovation_rms": 0.23683305415728928, + "instruction_rms": 0.08057459512777704, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1396.0635986328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 440109.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 416.70928955078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1828025579452515, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5946528315544128, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.85946655273438, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 106.79360961914062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 180.6107635498047, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1481.216064453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2259.204833984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 416.7093811035156, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.187349557876587, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5947977304458618, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.7889862060547, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3600993399183776, + "step": 65, + "teaching_rms": 0.23683305415728928, + "traffic_rms": 0.35096911215355187 + }, + { + "batch_loss": 2.564851999282837, + "innovation_rms": 0.31164300753587487, + "instruction_rms": 0.011725671278669615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1563.8875732421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 711740.25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 644.6356201171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1742421388626099, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36937159299850464, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.79757690429688, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 131.64797973632812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 248.54373168945312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1773.3336181640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2921.17529296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 644.6356201171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.178788661956787, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3695164918899536, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.7270965576172, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.43658183533522293, + "step": 66, + "teaching_rms": 0.31164300753587487, + "traffic_rms": 0.43642448870155437 + }, + { + "batch_loss": 2.9277164936065674, + "innovation_rms": 0.12901590503433677, + "instruction_rms": 0.012726671557968573, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1305.14013671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 835252.8125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 575.3596801757812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.150208830833435, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5136908292770386, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 259.9652099609375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.36927795410156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.21295928955078, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2035.3824462890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2620.48828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 575.3597412109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.154754877090454, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5133702158927917, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 259.91082763671875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.34342994680669703, + "step": 67, + "teaching_rms": 0.12901590503433677, + "traffic_rms": 0.34319461112315824 + }, + { + "batch_loss": 2.7110092639923096, + "innovation_rms": 0.11731177767800663, + "instruction_rms": 0.06590872812180482, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 994.9151611328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1122645.25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 559.735107421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1411356925964355, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41180431842803955, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 315.938720703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 142.0532989501953, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.15973663330078, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2257.914794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2225.32275390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 559.7351684570312, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1456812620162964, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41191309690475464, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 315.88433837890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.4280361332968367, + "step": 68, + "teaching_rms": 0.11731177767800663, + "traffic_rms": 0.4229316077624109 + }, + { + "batch_loss": 2.6919002532958984, + "innovation_rms": 0.311625413753151, + "instruction_rms": 0.2882152622464638, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 846.1014404296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1451099.75, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541.810546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.16278874874115, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.701188325881958, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 370.1197814941406, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 126.68783569335938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.65460205078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2446.888427734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889.736328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541.8106079101562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1673338413238525, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7012971639633179, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 370.0653991699219, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.5367136747137968, + "step": 69, + "teaching_rms": 0.311625413753151, + "traffic_rms": 0.4527622472226373 + }, + { + "batch_loss": 2.683516263961792, + "innovation_rms": 0.22311774942885151, + "instruction_rms": 0.0660014415509109, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 975.3611450195312, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1699930.625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 736.3604736328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1743135452270508, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38738229870796204, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.9798889160156, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 105.2354507446289, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.52381896972656, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2627.25048828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1803.62060546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 736.3605346679688, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1788581609725952, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.387515127658844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.9255065917969, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.4765346005592224, + "step": 70, + "teaching_rms": 0.22311774942885151, + "traffic_rms": 0.47194173336348644 + }, + { + "batch_loss": 2.88777494430542, + "innovation_rms": 0.49723513059474794, + "instruction_rms": 0.054381296179214736, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1162.4237060546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2412714.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1481.525390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.164742350578308, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7209041714668274, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 431.9229431152344, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.66265106201172, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.72799682617188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2906.490966796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2792.404296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1481.525390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1692866086959839, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7207953929901123, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 431.8685607910156, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7436776155983003, + "step": 71, + "teaching_rms": 0.49723513059474794, + "traffic_rms": 0.7416866945711909 + }, + { + "batch_loss": 2.7985482215881348, + "innovation_rms": 0.5069711965763046, + "instruction_rms": 0.02103685515817409, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1643.0394287109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3597376.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 938.61572265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.135682463645935, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5232484936714172, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 485.0625915527344, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.93904876708984, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.76397705078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3262.56396484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3560.728759765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 938.61572265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1402262449264526, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5233424305915833, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 485.00823974609375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8930689405778567, + "step": 72, + "teaching_rms": 0.5069711965763046, + "traffic_rms": 0.8928212457743013 + }, + { + "batch_loss": 2.6678414344787598, + "innovation_rms": 0.5959988062709746, + "instruction_rms": 0.0056863490005857826, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2120.9765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4855907.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1609.1025390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1163495779037476, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5835896134376526, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 645.5194091796875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.0494384765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.10389709472656, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3720.547607421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4579.8369140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1609.1025390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1006327867507935, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5834567546844482, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 645.5280151367188, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9622977698112016, + "step": 73, + "teaching_rms": 0.5959988062709746, + "traffic_rms": 0.9622810412972747 + }, + { + "batch_loss": 2.5630953311920166, + "innovation_rms": 0.6901401666902398, + "instruction_rms": 0.024318545207142165, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2793.4609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5895401.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1759.958740234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1320587396621704, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.478259414434433, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 732.5122680664062, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.11021423339844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 310.6077880859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4302.12744140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5815.79638671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1759.958740234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.067610740661621, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4780871868133545, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 732.5208740234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9734595800197348, + "step": 74, + "teaching_rms": 0.6901401666902398, + "traffic_rms": 0.973155823294966 + }, + { + "batch_loss": 2.5907602310180664, + "innovation_rms": 0.9369858055135685, + "instruction_rms": 0.018531792159115316, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3593.6318359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7748803.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1705.7100830078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.138120412826538, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30350011587142944, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 825.2273559570312, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 204.67347717285156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 575.6326293945312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5111.5654296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8094.3779296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1705.7100830078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0669828653335571, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.303387314081192, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 825.2359619140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.2012941359768132, + "step": 75, + "teaching_rms": 0.9369858055135685, + "traffic_rms": 1.2011512222015557 + }, + { + "batch_loss": 2.5650691986083984, + "innovation_rms": 0.9331135419132125, + "instruction_rms": 0.007899093330351166, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3954.60205078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9463860.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2658.0849609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1536518335342407, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5123666524887085, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 980.4601440429688, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 274.0263977050781, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 693.5291748046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6071.9072265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9603.41796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2658.085205078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0631496906280518, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.512526273727417, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 980.46875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.2666657814697966, + "step": 76, + "teaching_rms": 0.9331135419132125, + "traffic_rms": 1.2666412033899417 + }, + { + "batch_loss": 2.474125862121582, + "innovation_rms": 0.8986322104740131, + "instruction_rms": 0.02285929180078884, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6155.033203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12920832.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1776.378662109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.192914366722107, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4762081205844879, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1138.751220703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 336.7297058105469, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 627.032958984375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7073.86279296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10019.5546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1776.378662109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.060442566871643, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4763677418231964, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1138.7598876953125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.6037263563398982, + "step": 77, + "teaching_rms": 0.8986322104740131, + "traffic_rms": 1.6035634280712907 + }, + { + "batch_loss": 2.766940116882324, + "innovation_rms": 0.9660500106823019, + "instruction_rms": 0.04215730618124908, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9126.4052734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17395740.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2178.343017578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2023004293441772, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5079782009124756, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1305.6983642578125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 385.1786193847656, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 484.4891357421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8096.2646484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10224.01953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2178.343017578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0661022663116455, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5080909729003906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1305.70703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.8546986399966847, + "step": 78, + "teaching_rms": 0.9660500106823019, + "traffic_rms": 1.8542194638643927 + }, + { + "batch_loss": 2.5057625770568848, + "innovation_rms": 1.2204885803735013, + "instruction_rms": 0.015047596895545183, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12064.4580078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24590448.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2635.80908203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1991353034973145, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5108559131622314, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1499.6605224609375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 428.6344909667969, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 434.55865478515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9210.6240234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11143.58984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2635.80908203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.062853217124939, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5110607743263245, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1499.669189453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.1669538210531085, + "step": 79, + "teaching_rms": 1.2204885803735013, + "traffic_rms": 2.1669015729521197 + }, + { + "batch_loss": 2.6486597061157227, + "innovation_rms": 1.5814475330401088, + "instruction_rms": 0.09548419596560503, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13943.478515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34032764.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4789.9697265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2042156457901, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6315016746520996, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1757.1917724609375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 485.41204833984375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 567.7756958007812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10530.6171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13199.935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4789.9697265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0408507585525513, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6313889026641846, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1757.200439453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.626749705590453, + "step": 80, + "teaching_rms": 1.5814475330401088, + "traffic_rms": 2.6250137247132304 + }, + { + "batch_loss": 2.3567957878112793, + "innovation_rms": 1.7934331271398425, + "instruction_rms": 0.062078883027100226, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13465.4736328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 40928520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5013.529296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2126015424728394, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40827643871307373, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2014.78076171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 562.94482421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 775.3277587890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12118.8701171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15882.5244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5013.529296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0307551622390747, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4081636667251587, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2014.7894287109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.5137616564096876, + "step": 81, + "teaching_rms": 1.7934331271398425, + "traffic_rms": 2.512995018301379 + }, + { + "batch_loss": 2.5931732654571533, + "innovation_rms": 1.8455166882201968, + "instruction_rms": 0.0275920675708542, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17028.7734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 54994608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3688.41259765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2186397314071655, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43427979946136475, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2290.197265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 634.2454223632812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 713.0059204101562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13848.0859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17292.154296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3688.41259765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0562362670898438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43431413173675537, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2290.2060546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.218825119673509, + "step": 82, + "teaching_rms": 1.8455166882201968, + "traffic_rms": 3.2187068536050893 + }, + { + "batch_loss": 2.9114420413970947, + "innovation_rms": 1.9477108722791987, + "instruction_rms": 0.12357752947523508, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21993.4375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73442088.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4304.357421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2225544452667236, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6787253022193909, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2578.307373046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 686.2507934570312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 520.053955078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15666.6962890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18186.107421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4304.357421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0746501684188843, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6786982417106628, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2578.316162109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.6334470187954637, + "step": 83, + "teaching_rms": 1.9477108722791987, + "traffic_rms": 3.6313449111654443 + }, + { + "batch_loss": 2.615722417831421, + "innovation_rms": 2.4308334518247614, + "instruction_rms": 0.04157323506658644, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26772.755859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 103159592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4503.525390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2261089086532593, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49284911155700684, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2913.008544921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 736.318359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 500.67535400390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17719.1171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20524.212890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4503.525390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.062441349029541, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4926351308822632, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2913.017333984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.1998719226151975, + "step": 84, + "teaching_rms": 2.4308334518247614, + "traffic_rms": 4.199666180308627 + }, + { + "batch_loss": 3.027332305908203, + "innovation_rms": 3.3589303705497167, + "instruction_rms": 0.13175662553591774, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29609.796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 137953632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5893.384765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2293510437011719, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.978291392326355, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3426.423583984375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 848.3775024414062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1120.59130859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20380.685546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26615.6796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5893.384765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0268675088882446, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9784379005432129, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3426.432373046875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.0816989048025, + "step": 85, + "teaching_rms": 3.3589303705497167, + "traffic_rms": 5.079990562811194 + }, + { + "batch_loss": 2.981295347213745, + "innovation_rms": 3.2969547450945846, + "instruction_rms": 0.013782826565211497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 28382.291015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 158722992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7709.556640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2323791980743408, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8663213849067688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4003.778076171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 984.2557983398438, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1358.783203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23461.33984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30806.53515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7709.556640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9890250563621521, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8665199875831604, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4003.786865234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.758571479788445, + "step": 86, + "teaching_rms": 3.2969547450945846, + "traffic_rms": 4.758551548945416 + }, + { + "batch_loss": 3.2111639976501465, + "innovation_rms": 4.901612549118627, + "instruction_rms": 0.05055850845593573, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 30915.275390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 232109328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7182.76318359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2352887392044067, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1398338079452515, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4432.71923828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1250.3021240234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2660.462890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 27625.638671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41642.9921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7182.76318359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0923678874969482, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1397523880004883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4432.72802734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.646987482737883, + "step": 87, + "teaching_rms": 4.901612549118627, + "traffic_rms": 6.646795222606981 + }, + { + "batch_loss": 3.0289487838745117, + "innovation_rms": 5.887815518004567, + "instruction_rms": 0.009915300413684203, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34536.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 315397568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9392.796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.237768292427063, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7911112308502197, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4933.0361328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1663.5152587890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4132.1318359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 33116.33203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54906.93359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9392.796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1626100540161133, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7910298109054565, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4933.044921875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.635853904494076, + "step": 88, + "teaching_rms": 5.887815518004567, + "traffic_rms": 7.635847482274969 + }, + { + "batch_loss": 3.605638265609741, + "innovation_rms": 6.551307961503503, + "instruction_rms": 0.24420775446916132, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36446.88671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 393605632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12309.41796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2393265962600708, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.569900631904602, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5640.14892578125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2208.85888671875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5453.43603515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 39865.140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67488.1015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12309.41796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1775425672531128, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5699295997619629, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5640.1650390625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8.470810126128148, + "step": 89, + "teaching_rms": 6.551307961503503, + "traffic_rms": 8.467289244633154 + }, + { + "batch_loss": 3.0846385955810547, + "innovation_rms": 7.117811497906396, + "instruction_rms": 0.020939364666174718, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 46501.58203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 556572800.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12062.044921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2401031255722046, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6615093350410461, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6742.40869140625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2774.4951171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5656.3623046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 47536.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76716.09375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12062.044921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1608370542526245, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6615383625030518, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6742.4248046875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11.214017447418323, + "step": 90, + "teaching_rms": 7.117811497906396, + "traffic_rms": 11.213997907752432 + }, + { + "batch_loss": 2.7253620624542236, + "innovation_rms": 7.903008611611424, + "instruction_rms": 0.096101390605909, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56616.3984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 734328640.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14995.02734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2407326698303223, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46995601058006287, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8241.9111328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3370.286865234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5957.9169921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 56100.48828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85637.3984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14995.02734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1198810338974, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46998506784439087, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8241.927734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12.17793342974747, + "step": 91, + "teaching_rms": 7.903008611611424, + "traffic_rms": 12.177554239677573 + }, + { + "batch_loss": 2.9531383514404297, + "innovation_rms": 11.392468824122318, + "instruction_rms": 1.4902839060801483, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 60363.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1016573888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41083.93359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2420693635940552, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7920073866844177, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9054.0400390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4202.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8323.380859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 66942.3828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108418.9609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41083.93359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0541988611221313, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7918230295181274, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9054.056640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15.09439533422483, + "step": 92, + "teaching_rms": 11.392468824122318, + "traffic_rms": 15.020646606512098 + }, + { + "batch_loss": 2.5661420822143555, + "innovation_rms": 12.045317469762441, + "instruction_rms": 0.7502247595709868, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 73096.3984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1653744896.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38966.21875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2438013553619385, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5715024471282959, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10603.904296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5135.8251953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9332.001953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 79284.234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123418.546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38966.21875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0290112495422363, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5713180899620056, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10603.9208984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20.21337703640122, + "step": 93, + "teaching_rms": 12.045317469762441, + "traffic_rms": 20.199449843544492 + }, + { + "batch_loss": 2.6598050594329834, + "innovation_rms": 14.368176816272697, + "instruction_rms": 0.5450545014147837, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 88803.3828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2483663360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63169.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2455008029937744, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34767842292785645, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15079.345703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6191.384765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10555.595703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 93549.859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142656.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63169.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0407719612121582, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3474952280521393, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15079.431640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23.42282741673072, + "step": 94, + "teaching_rms": 14.368176816272697, + "traffic_rms": 23.416484789409076 + }, + { + "batch_loss": 2.915207624435425, + "innovation_rms": 20.472754024696453, + "instruction_rms": 0.7617634350210508, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 100301.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3450461696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76501.1015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2474591732025146, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5926806926727295, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22729.455078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7793.4384765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16020.53515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 112688.5703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191387.125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76501.1015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0346332788467407, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5923617482185364, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22729.541015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26.179454728404107, + "step": 95, + "teaching_rms": 20.472754024696453, + "traffic_rms": 26.168369587098105 + }, + { + "batch_loss": 3.017101764678955, + "innovation_rms": 9.00437041954693, + "instruction_rms": 0.5054029967488326, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 95379.5625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3939945216.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53066.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2483724355697632, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.699354350566864, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28036.111328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705.994140625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9125.5556640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 130186.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 174974.265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53066.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0158299207687378, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6993163824081421, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28036.197265625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20.895911519719785, + "step": 96, + "teaching_rms": 9.00437041954693, + "traffic_rms": 20.889798616715346 + }, + { + "batch_loss": 2.9514708518981934, + "innovation_rms": 11.432937077694667, + "instruction_rms": 0.5506219801015574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 93893.609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4641087488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34912.6015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2454643249511719, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6687343716621399, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30612.646484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9004.287109375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2982.9296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 146498.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163120.90625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34912.6015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9850057363510132, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6690537929534912, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30612.732421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 25.019672252021802, + "step": 97, + "teaching_rms": 11.432937077694667, + "traffic_rms": 25.013612596442236 + }, + { + "batch_loss": 2.939070701599121, + "innovation_rms": 30.845611102786833, + "instruction_rms": 0.5454424422865625, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 120036.5234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281275904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78466.6953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.243788480758667, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8707284331321716, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33268.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10072.048828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10677.61328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 170060.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 235619.71875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78466.6953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.957617461681366, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.870966374874115, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33268.8359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41.86190551346288, + "step": 98, + "teaching_rms": 30.845611102786833, + "traffic_rms": 41.85835192613411 + }, + { + "batch_loss": 3.0843405723571777, + "innovation_rms": 31.843570927450234, + "instruction_rms": 0.3954096550532135, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 153414.53125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12147561472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53353.3515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2455326318740845, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5681056976318359, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 37574.60546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11741.07421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16690.25390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 198624.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285648.6875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53353.3515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9421226978302002, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5683436393737793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 37574.69140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52.80773826095765, + "step": 99, + "teaching_rms": 31.843570927450234, + "traffic_rms": 52.80625788408374 + }, + { + "batch_loss": 2.925293207168579, + "innovation_rms": 43.12883298357078, + "instruction_rms": 0.3301051041996635, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 189384.640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 16263103488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66175.984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.252729058265686, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.679929792881012, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43831.15234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 14655.310546875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29142.365234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 237580.46875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 389555.3125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66175.984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9391531944274902, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.680103600025177, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43831.23828125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 54.61619068472097, + "step": 100, + "teaching_rms": 43.12883298357078, + "traffic_rms": 54.615193085721444 + }, + { + "batch_loss": 2.8266639709472656, + "innovation_rms": 43.82488468235351, + "instruction_rms": 0.49239893715795174, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 222474.421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19405441024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155023.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2432429790496826, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5547952651977539, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41449.87109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18243.478515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35881.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 283786.65625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 462061.90625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155023.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9470289945602417, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5548332333564758, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41449.95703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 58.23632937951089, + "step": 101, + "teaching_rms": 43.82488468235351, + "traffic_rms": 58.234247680066915 + }, + { + "batch_loss": 3.662792205810547, + "innovation_rms": 41.05551177473504, + "instruction_rms": 0.26391551913514294, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 317712.34375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 32044472320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123698.859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2395730018615723, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37808501720428467, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 49024.05859375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21442.548828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31990.701171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 330650.875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 468642.0625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123698.859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9466376304626465, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3781552314758301, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 49023.9921875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 88.2200018281759, + "step": 102, + "teaching_rms": 41.05551177473504, + "traffic_rms": 88.21960707019826 + }, + { + "batch_loss": 3.6556308269500732, + "innovation_rms": 41.696474343330316, + "instruction_rms": 0.4467335927147005, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 450994.84375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64248741888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103541.3828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2252906560897827, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6527115106582642, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 59378.1953125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 23834.080078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23915.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 376003.96875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 453531.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103541.3828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9355630874633789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6527817249298096, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 59378.12890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 104.65440279118381, + "step": 103, + "teaching_rms": 41.696474343330316, + "traffic_rms": 104.65344931203877 + }, + { + "batch_loss": 3.1498186588287354, + "innovation_rms": 45.50528795083991, + "instruction_rms": 0.4140605477513846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 607876.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 129895030784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86757.484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1837741136550903, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.977409839630127, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 68053.9453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25272.814453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14387.3369140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 419638.21875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 436342.46875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86757.484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9258806705474854, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9774856567382812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 68053.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 118.53876064101895, + "step": 104, + "teaching_rms": 45.50528795083991, + "traffic_rms": 118.53803747396184 + }, + { + "batch_loss": 3.0283455848693848, + "innovation_rms": 51.43574998891074, + "instruction_rms": 0.4793589208688736, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 781005.0625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 210207916032.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80793.7109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1631567478179932, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8725119829177856, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 75072.8359375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25833.951171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5611.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 462311.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 426729.15625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80793.7109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9444546103477478, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8726747035980225, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 75072.765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 132.2762268494219, + "step": 105, + "teaching_rms": 51.43574998891074, + "traffic_rms": 132.27535826775366 + }, + { + "batch_loss": 3.4575417041778564, + "innovation_rms": 57.23377795047765, + "instruction_rms": 1.3989412439348492, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 962211.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 297896738816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87820.9453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1588451862335205, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6345753073692322, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80473.8046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25579.2890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2546.6181640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 504757.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 424461.8125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87820.9453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9483126401901245, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6343911290168762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80473.734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 144.67875957883524, + "step": 106, + "teaching_rms": 57.23377795047765, + "traffic_rms": 144.67199603762444 + }, + { + "batch_loss": 2.9369454383850098, + "innovation_rms": 62.89025803453876, + "instruction_rms": 6.429681223701708, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1134753.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 379065008128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86952.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.162691354751587, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6942891478538513, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 84296.703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 24519.44921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10598.4052734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 547533.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 427763.4375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86952.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9460301995277405, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.694040060043335, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 84296.6328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 155.83307352062286, + "step": 107, + "teaching_rms": 62.89025803453876, + "traffic_rms": 155.700372517713 + }, + { + "batch_loss": 2.698108434677124, + "innovation_rms": 70.49683540573764, + "instruction_rms": 4.340719362257332, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1294861.625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 444540780544.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100176.7109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1683412790298462, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.525611162185669, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 86429.484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 22863.767578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16556.82421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 592327.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 447936.71875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100176.7109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9756701588630676, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5253621339797974, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 86429.4140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 165.3991697267064, + "step": 108, + "teaching_rms": 70.49683540573764, + "traffic_rms": 165.3422012126818 + }, + { + "batch_loss": 2.931814432144165, + "innovation_rms": 82.41276084519632, + "instruction_rms": 2.124608629272954, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1429593.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 480454836224.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155229.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.170103907585144, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.397666335105896, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 90750.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21110.74609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17530.20703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 642617.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 502901.625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155229.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0012820959091187, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3979412913322449, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 90750.765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 173.9956360833988, + "step": 109, + "teaching_rms": 82.41276084519632, + "traffic_rms": 173.9826641203834 + }, + { + "batch_loss": 3.071408748626709, + "innovation_rms": 107.2342055622969, + "instruction_rms": 1.4477689859761262, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1523230.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 478138105856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 261698.515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1578727960586548, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4557751417160034, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103605.5234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20751.646484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3591.00390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 708495.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 658775.75, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 261698.515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0109890699386597, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4560500383377075, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103605.53125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 181.57028933754336, + "step": 110, + "teaching_rms": 107.2342055622969, + "traffic_rms": 181.56451728037354 + }, + { + "batch_loss": 2.640561819076538, + "innovation_rms": 135.52137693599906, + "instruction_rms": 0.47361342308153753, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1524808.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 449764261888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 222424.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1476079225540161, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5703049898147583, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 123739.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 24056.05078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33044.03515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 803385.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 948899.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 222424.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0196739435195923, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5701068043708801, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 123739.234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 180.22101131323535, + "step": 111, + "teaching_rms": 135.52137693599906, + "traffic_rms": 180.22038899411436 + }, + { + "batch_loss": 2.9016144275665283, + "innovation_rms": 164.1159447457429, + "instruction_rms": 0.3125165117375963, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1525871.625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 433082105856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190438.265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1486350297927856, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5346590876579285, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 141060.515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32166.421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81103.703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 935709.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1323243.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190438.265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.019037127494812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5344641804695129, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 141060.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 214.04812002980754, + "step": 112, + "teaching_rms": 164.1159447457429, + "traffic_rms": 214.0478918888926 + }, + { + "batch_loss": 3.340075731277466, + "innovation_rms": 187.5714741794353, + "instruction_rms": 0.27244906874282765, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1664354.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 452221763584.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272124.34375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1572747230529785, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3947086334228516, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 166969.703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 44053.07421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118866.5234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1105689.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1699798.375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272124.34375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0144473314285278, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3943862915039062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 166969.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 255.50549494550614, + "step": 113, + "teaching_rms": 187.5714741794353, + "traffic_rms": 255.50534968848314 + }, + { + "batch_loss": 3.2119524478912354, + "innovation_rms": 228.85881485712105, + "instruction_rms": 0.26285197009796335, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1708019.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 541990289408.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 373658.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1767817735671997, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0319122076034546, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 174118.984375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 61732.9921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176799.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1325638.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2199494.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 373658.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0349206924438477, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0315899848937988, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 174118.84375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 289.877660825223, + "step": 114, + "teaching_rms": 228.85881485712105, + "traffic_rms": 289.8775416522569 + }, + { + "batch_loss": 3.2312405109405518, + "innovation_rms": 230.70748287098, + "instruction_rms": 0.2494260461961365, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1967793.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 768522977280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 327854.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.204316258430481, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6053696870803833, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 206394.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 80941.0390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192080.453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1570637.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2449984.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 327854.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.049362301826477, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6050474643707275, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 206394.796875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 375.5269591130475, + "step": 115, + "teaching_rms": 230.70748287098, + "traffic_rms": 375.5268762788022 + }, + { + "batch_loss": 3.0326597690582275, + "innovation_rms": 287.73571070848885, + "instruction_rms": 0.2262082878100995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2254399.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1103229878272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 570543.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.203927755355835, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7200848460197449, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 263449.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 103210.890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 222698.53125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855630.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2849937.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 570543.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.053753137588501, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7199393510818481, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 263449.15625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 454.4980115453296, + "step": 116, + "teaching_rms": 287.73571070848885, + "traffic_rms": 454.49795525257156 + }, + { + "batch_loss": 2.943819046020508, + "innovation_rms": 323.0424756820146, + "instruction_rms": 0.411525512279398, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2355734.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1293031309312.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1056063.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.155564546585083, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5260601043701172, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 290037.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 128927.3359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257164.46875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2187669.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3320388.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1056063.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.047479510307312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5260794162750244, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 290037.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 437.8298208045258, + "step": 117, + "teaching_rms": 323.0424756820146, + "traffic_rms": 437.82962740391855 + }, + { + "batch_loss": 3.189453601837158, + "innovation_rms": 323.7121120374936, + "instruction_rms": 1.0879591204593757, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3173176.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847120101376.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 716080.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1220262050628662, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0477097034454346, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 335718.875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 151619.796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 226924.671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2532467.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3447982.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 716080.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0204190015792847, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0476012229919434, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 335718.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 608.2618581059389, + "step": 118, + "teaching_rms": 323.7121120374936, + "traffic_rms": 608.2608851240227 + }, + { + "batch_loss": 3.4490089416503906, + "innovation_rms": 330.93853607215163, + "instruction_rms": 1.8764830120710352, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4396273.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3617495973888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 782625.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.12716805934906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8706505298614502, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 390660.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 168305.171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 166853.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2874701.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3422340.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 782625.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9737936854362488, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8706313371658325, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 390659.90625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 698.8718545176064, + "step": 119, + "teaching_rms": 330.93853607215163, + "traffic_rms": 698.8693353189916 + }, + { + "batch_loss": 3.0174272060394287, + "innovation_rms": 349.0306058289314, + "instruction_rms": 0.12639401672740527, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5825698.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7044931256320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 913711.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1585121154785156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4215455055236816, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 449716.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 178124.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98188.5703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3210690.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3359885.75, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 913711.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9850819706916809, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4215261936187744, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 449715.90625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 770.6509492526926, + "step": 120, + "teaching_rms": 349.0306058289314, + "traffic_rms": 770.6509388877591 + }, + { + "batch_loss": 3.795043468475342, + "innovation_rms": 416.0778064550746, + "instruction_rms": 1.0538432649394032, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7379731.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11106994094080.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1031365.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2091517448425293, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0361628532409668, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 517440.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 181919.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37951.32421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3553274.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3425839.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1031365.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9943071007728577, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0361247062683105, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 517440.96875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 880.6504894230608, + "step": 121, + "teaching_rms": 416.0778064550746, + "traffic_rms": 880.6498588744729 + }, + { + "batch_loss": 4.717951774597168, + "innovation_rms": 478.8642224296497, + "instruction_rms": 1.2283072140896445, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8854510.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14881072873472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1130369.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2648406028747559, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4654737710952759, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 630477.875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 181420.609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4985.4140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3915157.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3618826.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1130369.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.98363196849823, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4654459953308105, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 630477.9375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 973.5306587589234, + "step": 122, + "teaching_rms": 478.8642224296497, + "traffic_rms": 973.5298838793262 + }, + { + "batch_loss": 3.527984619140625, + "innovation_rms": 584.5586153300495, + "instruction_rms": 1.069727660280836, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10038373.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17184561561600.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1182384.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.317795753479004, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9430664777755737, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 748716.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 181969.734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5491.25244140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4328161.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4130039.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1182384.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0049411058425903, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9430387020111084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 748716.3125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1089.5535248308906, + "step": 123, + "teaching_rms": 584.5586153300495, + "traffic_rms": 1089.5529996996943 + }, + { + "batch_loss": 4.550325393676758, + "innovation_rms": 782.1404094351735, + "instruction_rms": 0.5618852669652941, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10676231.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17416732016640.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2499392.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2618032693862915, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8488470315933228, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 847915.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 195406.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 134367.296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4872857.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5446959.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2499392.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.948979914188385, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8488138914108276, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 847915.3125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1254.4493207571018, + "step": 124, + "teaching_rms": 782.1404094351735, + "traffic_rms": 1254.4491949190906 + }, + { + "batch_loss": 4.378256797790527, + "innovation_rms": 697.7643735969441, + "instruction_rms": 0.40563086193188735, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10310451.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 16863341838336.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2234585.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2266045808792114, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0241602659225464, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 954818.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 208659.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132528.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5485332.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6124755.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2234585.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.913812518119812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0240211486816406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 954818.3125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1041.756789294219, + "step": 125, + "teaching_rms": 697.7643735969441, + "traffic_rms": 1041.7567103239635 + }, + { + "batch_loss": 4.766293048858643, + "innovation_rms": 1084.1148705143933, + "instruction_rms": 0.3097509164050097, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11065359.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 16917727281152.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1703530.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2042546272277832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9427011609077454, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1079797.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 246598.890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 379395.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6322131.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8367986.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1703530.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9174820184707642, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9425619840621948, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1079797.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1504.625826769009, + "step": 126, + "teaching_rms": 1084.1148705143933, + "traffic_rms": 1504.625794885595 + }, + { + "batch_loss": 3.576258897781372, + "innovation_rms": 1299.6611165013762, + "instruction_rms": 0.24742918860958846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11743141.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19346441109504.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2321468.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1854966878890991, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0059874057769775, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1157776.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 317112.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 705140.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7436350.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11142193.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2321468.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9541501998901367, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0058287382125854, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1157776.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1828.0269407286098, + "step": 127, + "teaching_rms": 1299.6611165013762, + "traffic_rms": 1828.0269239835095 + }, + { + "batch_loss": 3.1089963912963867, + "innovation_rms": 1427.6747052630135, + "instruction_rms": 0.32790850260104637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12040637.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22493538549760.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3327499.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.168602705001831, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9882718324661255, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1393534.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 418353.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1012406.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822716.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13863656.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3327499.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9794673919677734, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9882419109344482, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1393534.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1837.4939748333213, + "step": 128, + "teaching_rms": 1427.6747052630135, + "traffic_rms": 1837.4939455750548 + }, + { + "batch_loss": 2.863762378692627, + "innovation_rms": 1512.5866656318385, + "instruction_rms": 0.40021730477894774, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14303887.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 32226483896320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2414631.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1533863544464111, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9787696599960327, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1564162.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 526387.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1080341.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10377717.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15550005.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2414631.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.996431827545166, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9790319204330444, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1564162.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2511.889118334292, + "step": 129, + "teaching_rms": 1512.5866656318385, + "traffic_rms": 2511.88908645111 + }, + { + "batch_loss": 3.2466955184936523, + "innovation_rms": 1687.86703147428, + "instruction_rms": 0.7122092345113342, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17127094.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 45873337729024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2437092.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1396801471710205, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7392107248306274, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1672925.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 635631.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1092433.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12085590.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17078734.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2437092.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0023584365844727, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7394729256629944, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1672925.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2964.1203671143758, + "step": 130, + "teaching_rms": 1687.86703147428, + "traffic_rms": 2964.120281550778 + }, + { + "batch_loss": 3.177769184112549, + "innovation_rms": 2257.9336797587443, + "instruction_rms": 0.7086531163800945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19559270.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 65252901257216.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4957898.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.127333164215088, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8873834609985352, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730434.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 775695.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1400644.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14157098.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20715082.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4957898.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.00437593460083, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8872808218002319, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730434.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3468.6123246526445, + "step": 131, + "teaching_rms": 2257.9336797587443, + "traffic_rms": 3468.6122522622354 + }, + { + "batch_loss": 2.7658350467681885, + "innovation_rms": 2062.28536588471, + "instruction_rms": 0.6392194292494019, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19992560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72288292569088.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6054416.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.116209626197815, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6252777576446533, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2020140.125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 904197.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1285022.625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16394130.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22370316.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6054416.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.009324550628662, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6251751184463501, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2020140.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3098.636399944063, + "step": 132, + "teaching_rms": 2062.28536588471, + "traffic_rms": 3098.6363340116172 + }, + { + "batch_loss": 3.1688928604125977, + "innovation_rms": 3126.0051665997885, + "instruction_rms": 0.3573416802369595, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24216718.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 103930885308416.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5575075.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1061872243881226, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5565117597579956, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2556467.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1079347.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1751494.125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19170912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27767822.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5575075.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0138046741485596, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5564819574356079, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2556467.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4611.473493352158, + "step": 133, + "teaching_rms": 3126.0051665997885, + "traffic_rms": 4611.473479507024 + }, + { + "batch_loss": 3.387921094894409, + "innovation_rms": 3881.1557399696976, + "instruction_rms": 0.3725153147314392, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 28579254.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 157574666125312.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560228.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.097156047821045, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8782595992088318, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3108192.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1330345.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2509981.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22710686.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35397740.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560228.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0150551795959473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8781499862670898, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3108192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5616.80560303352, + "step": 134, + "teaching_rms": 3881.1557399696976, + "traffic_rms": 5616.805590680659 + }, + { + "batch_loss": 2.7914018630981445, + "innovation_rms": 3400.7590086641953, + "instruction_rms": 0.30963927953255643, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29327394.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 175685100371968.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9676144.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0890170335769653, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7301984429359436, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3661424.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1574226.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2438807.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 26549712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38390268.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9676144.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0170533657073975, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7300887703895569, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3661424.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4921.459248980768, + "step": 135, + "teaching_rms": 3400.7590086641953, + "traffic_rms": 4921.459239240113 + }, + { + "batch_loss": 3.1908466815948486, + "innovation_rms": 4760.608532769913, + "instruction_rms": 0.45932801060772327, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37841524.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 259857550671872.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8497278.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0816810131072998, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8392083644866943, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4511152.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853824.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2795986.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 31052300.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45025888.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8497278.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0156853199005127, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.83940589427948, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4511152.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7409.732932032644, + "step": 136, + "teaching_rms": 4760.608532769913, + "traffic_rms": 7409.732917795835 + }, + { + "batch_loss": 2.8859994411468506, + "innovation_rms": 5629.8612713798, + "instruction_rms": 0.44254681816183145, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 47331004.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 384756759920640.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10702314.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.075063705444336, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.975231409072876, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5581384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2185859.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3320350.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 36358304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53060048.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10702314.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0181710720062256, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.975412130355835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5581384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8886.91874147676, + "step": 137, + "teaching_rms": 5629.8612713798, + "traffic_rms": 8886.91873045789 + }, + { + "batch_loss": 3.1533501148223877, + "innovation_rms": 7943.879345153944, + "instruction_rms": 0.5174531458391697, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 51802568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 564704112017408.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12196251.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.069207787513733, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9082887768745422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6244521.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2754732.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5688725.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 43568184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72098816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12196251.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0146607160568237, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.908076286315918, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6244521.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10313.70270912099, + "step": 138, + "teaching_rms": 7943.879345153944, + "traffic_rms": 10313.702696140346 + }, + { + "batch_loss": 3.428781270980835, + "innovation_rms": 5413.832228992532, + "instruction_rms": 0.522727456516206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 54333440.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 637635139731456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12477298.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0645660161972046, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6204981207847595, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294934.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3249822.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4950906.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 50829032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72608472.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12477298.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0057693719863892, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6202856302261353, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294934.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9037.484851836896, + "step": 139, + "teaching_rms": 5413.832228992532, + "traffic_rms": 9037.484836719654 + }, + { + "batch_loss": 2.9520466327667236, + "innovation_rms": 10369.494271664567, + "instruction_rms": 0.5653726803226765, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 62715068.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 913062936707072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27802752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0625202655792236, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7831572890281677, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9653736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3961157.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7113347.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 60231900.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94028664.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27802752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.000232458114624, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7830610871315002, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9653736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14212.013545424248, + "step": 140, + "teaching_rms": 10369.494271664567, + "traffic_rms": 14212.01353417862 + }, + { + "batch_loss": 2.7612345218658447, + "innovation_rms": 8232.069155025027, + "instruction_rms": 0.6972479393764787, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 68187848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1431808012976128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21981382.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0595561265945435, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6589882969856262, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11472180.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4639994.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6788364.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 69792560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95606576.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21981382.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.003562331199646, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6587651968002319, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11472180.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17252.212131860073, + "step": 141, + "teaching_rms": 8232.069155025027, + "traffic_rms": 17252.212117770443 + }, + { + "batch_loss": 2.8146321773529053, + "innovation_rms": 9102.237495428895, + "instruction_rms": 0.6831310431935164, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 76399864.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2150636825083904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20845816.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.053944706916809, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9093703627586365, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12914483.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5278476.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6384822.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 79450896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 96583384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20845816.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.996519148349762, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9093953967094421, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12914483.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20314.47457691821, + "step": 142, + "teaching_rms": 9102.237495428895, + "traffic_rms": 20314.474565432116 + }, + { + "batch_loss": 2.9687376022338867, + "innovation_rms": 12658.701427103299, + "instruction_rms": 0.6551643283751594, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 90758120.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3055337997336576.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34892384.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.050175428390503, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6933075785636902, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14221451.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6022324.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7438480.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 90447688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109967912.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34892384.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9885625839233398, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6929895877838135, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14221451.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22945.34373327101, + "step": 143, + "teaching_rms": 12658.701427103299, + "traffic_rms": 22945.34372391747 + }, + { + "batch_loss": 2.936171293258667, + "innovation_rms": 17835.417787676895, + "instruction_rms": 0.5971872270554114, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 112889512.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4021335734878208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32973472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0473620891571045, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.543472170829773, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17257334.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7144984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11226602.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 104850560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144028704.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32973472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0109281539916992, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.543350100517273, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17257334.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26012.84756172801, + "step": 144, + "teaching_rms": 17835.417787676895, + "traffic_rms": 26012.847554873082 + }, + { + "batch_loss": 2.702181816101074, + "innovation_rms": 18730.864488898456, + "instruction_rms": 0.5723295864178028, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 132201936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4569380809605120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71005016.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0436909198760986, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6382465362548828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21707718.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8568626.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14236424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 122373280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 175227168.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71005016.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0310471057891846, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6381244659423828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21707718.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 25075.889718954124, + "step": 145, + "teaching_rms": 18730.864488898456, + "traffic_rms": 25075.88971242273 + }, + { + "batch_loss": 2.699154853820801, + "innovation_rms": 23559.18038976243, + "instruction_rms": 0.6807716092635292, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 154845504.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7455964978479104.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56437672.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0391085147857666, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4066677391529083, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25022714.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10450150.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18815236.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 143781152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214078720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56437672.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0491437911987305, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4065456986427307, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25022714.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 38239.80033299012, + "step": 146, + "teaching_rms": 23559.18038976243, + "traffic_rms": 38239.80032693034 + }, + { + "batch_loss": 2.5505611896514893, + "innovation_rms": 30637.864114983713, + "instruction_rms": 0.5575046339387586, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 181237024.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0540588478758912e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113208016.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0345971584320068, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49642443656921387, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28541990.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13057035.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26068848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 171932896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 281517504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113208016.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.065420389175415, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4962425231933594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28541990.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41947.326510211744, + "step": 147, + "teaching_rms": 30637.864114983713, + "traffic_rms": 41947.32650650696 + }, + { + "batch_loss": 2.550424337387085, + "innovation_rms": 23805.245288110378, + "instruction_rms": 0.5201724330862664, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 205583360.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1391016699428864e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65665544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.03141188621521, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6073606014251709, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33308798.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15564548.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25075132.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 201434000.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 295011104.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65665544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0800586938858032, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6071771383285522, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 33308798.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35921.222254686734, + "step": 148, + "teaching_rms": 23805.245288110378, + "traffic_rms": 35921.22225092046 + }, + { + "batch_loss": 2.9406521320343018, + "innovation_rms": 43442.77137778443, + "instruction_rms": 0.5355287339562754, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 252593952.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.570086498336768e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135887040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0282312631607056, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0814695358276367, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40509800.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19018956.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34544072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 240909696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 394756928.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135887040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0931788682937622, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0816149711608887, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40509800.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 56952.91485253286, + "step": 149, + "teaching_rms": 43442.77137778443, + "traffic_rms": 56952.91485001508 + }, + { + "batch_loss": 2.759784460067749, + "innovation_rms": 37712.069807045504, + "instruction_rms": 0.6601641956906482, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 275287648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5096069436145664e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122440680.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0379722118377686, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0910569429397583, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46235068.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 22630340.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36113844.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 282838400.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 419287136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122440680.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1040865182876587, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0912023782730103, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46235068.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 72786.79670344673, + "step": 150, + "teaching_rms": 37712.069807045504, + "traffic_rms": 72786.79670045295 + }, + { + "batch_loss": 2.796752452850342, + "innovation_rms": 47874.282212063765, + "instruction_rms": 0.7903619814445358, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 306323648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.805823349083341e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 276571648.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0572330951690674, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8778440952301025, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 71897952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 26739932.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41095928.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 329995936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 471575328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 276571648.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.112921953201294, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8775703310966492, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 71897952.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 85854.77229711761, + "step": 151, + "teaching_rms": 47874.282212063765, + "traffic_rms": 85854.77229347965 + }, + { + "batch_loss": 2.586143732070923, + "innovation_rms": 65319.618330886406, + "instruction_rms": 0.6766725559341822, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 377485952.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.860996740972544e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 148014224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0616217851638794, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.087349772453308, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 86040768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31907908.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51679764.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 389320224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 593242880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 148014224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1245537996292114, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0875331163406372, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 86040768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 93352.40969352824, + "step": 152, + "teaching_rms": 65319.618330886406, + "traffic_rms": 93352.40969107578 + }, + { + "batch_loss": 2.7055630683898926, + "innovation_rms": 77793.32118657764, + "instruction_rms": 0.6070106106221511, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 429817728.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.894343109063475e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 221098752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.163169503211975, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0154776573181152, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108150640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 38529632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66217248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 463760736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 744405120.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 221098752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1676771640777588, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0154731273651123, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108150640.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 102847.41906348773, + "step": 153, + "teaching_rms": 77793.32118657764, + "traffic_rms": 102847.41906169642 + }, + { + "batch_loss": 2.7074077129364014, + "innovation_rms": 74064.8392431188, + "instruction_rms": 0.5319253705928304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 548550016.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.373026717643571e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 205302560.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2284220457077026, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.861937940120697, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 128680896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 44966560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64369292.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 543491392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 797306304.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 205302560.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2329293489456177, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8617547154426575, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 128680896.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 130442.2534701464, + "step": 154, + "teaching_rms": 74064.8392431188, + "traffic_rms": 130442.25346906185 + }, + { + "batch_loss": 3.639218807220459, + "innovation_rms": 78027.43662162602, + "instruction_rms": 0.5920673178148144, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 700345216.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166827680998359e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239563424.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2534295320510864, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5401358604431152, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 147711024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 50673500.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57069400.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 625673792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 821824128.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239563424.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2579363584518433, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5402812957763672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 147711024.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 152162.66159722157, + "step": 155, + "teaching_rms": 78027.43662162602, + "traffic_rms": 152162.6615960697 + }, + { + "batch_loss": 2.8792483806610107, + "innovation_rms": 92480.86751815495, + "instruction_rms": 0.7501366273409257, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 870676736.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9704646811097498e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 329359680.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2549266815185547, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1331758499145508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 165294384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 55935572.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52620724.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 712994432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 873206400.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 329359680.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2594330310821533, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1333651542663574, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 165294384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 177740.0604024514, + "step": 156, + "teaching_rms": 92480.86751815495, + "traffic_rms": 177740.06040086847 + }, + { + "batch_loss": 4.163220405578613, + "innovation_rms": 120028.2538815274, + "instruction_rms": 0.6760951588768587, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1040421248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.585905147842396e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 418208832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2207814455032349, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9207028150558472, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 182971312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 62038384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61028104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 814660928.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1016664640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 418208832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2252873182296753, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9208730459213257, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 182971312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 209180.39634919175, + "step": 157, + "teaching_rms": 120028.2538815274, + "traffic_rms": 209180.39634809914 + }, + { + "batch_loss": 4.599479675292969, + "innovation_rms": 172622.12655340284, + "instruction_rms": 0.7174052492386327, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1167665664.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.327250691071672e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 446315520.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2628313302993774, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0646450519561768, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214450800.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 72638560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 106001744.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 955597504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1409365504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 446315520.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2673367261886597, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.064371347427368, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214450800.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 243809.009557449, + "step": 158, + "teaching_rms": 172622.12655340284, + "traffic_rms": 243809.00955639352 + }, + { + "batch_loss": 3.570925235748291, + "innovation_rms": 113449.6749449934, + "instruction_rms": 0.8813763439517646, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1220240384.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.581189086259446e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 366182464.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4304286241531372, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.67597234249115, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 251069040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 80575680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79371216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1096299520.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1407019520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 366182464.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4349335432052612, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.675967812538147, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 251069040.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 199146.48190649357, + "step": 159, + "teaching_rms": 113449.6749449934, + "traffic_rms": 199146.48190454318 + }, + { + "batch_loss": 4.348639011383057, + "innovation_rms": 179365.99715837569, + "instruction_rms": 1.289155181956319, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1353953408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3342927037569434e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 524834400.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5439229011535645, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.09785532951355, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 284995680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 90479960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99042800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1260532608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1642331264.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 524834400.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5484273433685303, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0977611541748047, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 284995680.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 265897.19737213774, + "step": 160, + "teaching_rms": 179365.99715837569, + "traffic_rms": 265897.1973690127 + }, + { + "batch_loss": 5.017866611480713, + "innovation_rms": 214704.22149254772, + "instruction_rms": 1.075081455871528, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1591414784.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.769332314963968e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 419435904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6133723258972168, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6132946014404297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 307182752.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 101950208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114702448.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1449952896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1894202624.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 419435904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6178762912750244, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6132004261016846, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 307182752.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 377355.64373669896, + "step": 161, + "teaching_rms": 214704.22149254772, + "traffic_rms": 377355.64373516745 + }, + { + "batch_loss": 5.641810417175293, + "innovation_rms": 227101.21708148337, + "instruction_rms": 0.765951304671334, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851379456.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0046905009108419e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 429646816.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6605955362319946, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3211666345596313, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 321366112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 113926464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119762552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1657462272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2075093888.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 429646816.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.665099024772644, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3210724592208862, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 321366112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 438654.6107240368, + "step": 162, + "teaching_rms": 227101.21708148337, + "traffic_rms": 438654.61072336807 + }, + { + "batch_loss": 5.3410773277282715, + "innovation_rms": 287818.4870332555, + "instruction_rms": 0.739642273835525, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2111778816.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3970631137521828e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 484209824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6989010572433472, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1403881311416626, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 331401664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 129006544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150800784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1903331968.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2458696704.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 484209824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.703404188156128, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1402939558029175, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 331401664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 489192.4018612799, + "step": 163, + "teaching_rms": 287818.4870332555, + "traffic_rms": 489192.40186072077 + }, + { + "batch_loss": 6.7927751541137695, + "innovation_rms": 411458.68370164325, + "instruction_rms": 1.068226689076736, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2333336576.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7411681969263084e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 792055232.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7326725721359253, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.53554105758667, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 363014720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 155879200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 268726592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2249161728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3458297856.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 792055232.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7371752262115479, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.535637855529785, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 363014720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 530436.9635501202, + "step": 164, + "teaching_rms": 411458.68370164325, + "traffic_rms": 530436.9635490446 + }, + { + "batch_loss": 4.422138214111328, + "innovation_rms": 455889.12914921314, + "instruction_rms": 0.7757927356215367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2547029504.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1351093318241485e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 966302208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7630314826965332, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7989866733551025, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 406000352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 192203344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 363241408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2688241408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4390797312.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 966302208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7675336599349976, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7990833520889282, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 406000352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 597876.3958556128, + "step": 165, + "teaching_rms": 455889.12914921314, + "traffic_rms": 597876.3958551096 + }, + { + "batch_loss": 3.5410966873168945, + "innovation_rms": 465693.96065000346, + "instruction_rms": 0.9184202144357839, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3080942336.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.290092008416215e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 788321152.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7903368473052979, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.397480845451355, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 461515616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 230852976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 386496352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3175646464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4874049536.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 788321152.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.794838547706604, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3975775241851807, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 461515616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 823789.799567072, + "step": 166, + "teaching_rms": 465693.96065000346, + "traffic_rms": 823789.79956656 + }, + { + "batch_loss": 3.9242472648620605, + "innovation_rms": 528349.9604818338, + "instruction_rms": 0.9906903298949374, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3704106496.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.771456053059518e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 934764928.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8148937225341797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2454994916915894, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 534828672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 270600864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 397478720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3710667776.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5350214144.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 934764928.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8193949460983276, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2455363273620605, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 534828672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 963988.4667920627, + "step": 167, + "teaching_rms": 528349.9604818338, + "traffic_rms": 963988.4667915538 + }, + { + "batch_loss": 3.4696388244628906, + "innovation_rms": 699437.5958149625, + "instruction_rms": 0.6770531889732131, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4229876992.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.612180354956526e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1719664128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8369767665863037, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.036548137664795, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 654951936.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 319363584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487627296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4356820992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6461531136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1719664128.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8414775133132935, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0362416505813599, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 654951936.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1092938.9003840208, + "step": 168, + "teaching_rms": 699437.5958149625, + "traffic_rms": 1092938.900383811 + }, + { + "batch_loss": 3.864351987838745, + "innovation_rms": 782836.476384801, + "instruction_rms": 0.5569073316531125, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4366181888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.646815848891417e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3108187136.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568331003189087, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7010958194732666, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 820532352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 376448832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 570852544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5127845888.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7710249984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3108187136.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8613334894180298, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7009638547897339, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 820532352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1064458.4129822347, + "step": 169, + "teaching_rms": 782836.476384801, + "traffic_rms": 1064458.4129820892 + }, + { + "batch_loss": 3.7031941413879395, + "innovation_rms": 826187.8521383369, + "instruction_rms": 0.6607697056298107, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5629796352.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.169922923053659e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2103923968.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.874685287475586, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5425182580947876, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1011561216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 434059136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 576103104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5966742528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8388965376.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2103923968.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8791851997375488, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5424242615699768, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1011561216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1539287.7657199271, + "step": 170, + "teaching_rms": 826187.8521383369, + "traffic_rms": 1539287.7657197854 + }, + { + "batch_loss": 3.3824687004089355, + "innovation_rms": 866715.7019456401, + "instruction_rms": 0.7176838178269197, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7370009600.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6854259810550415e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2533628928.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8907334804534912, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8646517992019653, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1175828352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 485805280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 517461440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6840427008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736843776.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2533628928.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.895232915878296, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8648638725280762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1175828352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1774049.4480587998, + "step": 171, + "teaching_rms": 866715.7019456401, + "traffic_rms": 1774049.4480586548 + }, + { + "batch_loss": 2.7276573181152344, + "innovation_rms": 1000846.9983585689, + "instruction_rms": 0.47602002757716144, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9182450688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3032798277070225e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3278541312.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9051579236984253, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0117155313491821, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1410176640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 531581664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 457763680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7766657024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9262302208.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3278541312.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9096570014953613, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0116872787475586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1410176640.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1980801.4645777275, + "step": 172, + "teaching_rms": 1000846.9983585689, + "traffic_rms": 1980801.4645776702 + }, + { + "batch_loss": 2.780879497528076, + "innovation_rms": 1276140.1648976463, + "instruction_rms": 0.5442700226342261, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10682336256.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0356805758383817e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3962405376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9181208610534668, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8837785124778748, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1783849344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 582819008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 512373504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8834797568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10681400320.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3962405376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9226195812225342, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8836311101913452, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1783849344.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2179844.8005258143, + "step": 173, + "teaching_rms": 1276140.1648976463, + "traffic_rms": 2179844.8005257463 + }, + { + "batch_loss": 3.2321841716766357, + "innovation_rms": 1837860.5406107956, + "instruction_rms": 0.5413722154492605, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11081073664.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.765541891914229e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3298635008.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9297683238983154, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9395666122436523, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868238848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 687963136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1051441216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10330745856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14959486976.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3298635008.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9342666864395142, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9397211074829102, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868238848.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2359442.5461733886, + "step": 174, + "teaching_rms": 1837860.5406107956, + "traffic_rms": 2359442.546173326 + }, + { + "batch_loss": 3.0960404872894287, + "innovation_rms": 1639131.3289531595, + "instruction_rms": 0.5215618101345679, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11132831744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.229489619347335e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7828737024.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9402318000793457, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6358999013900757, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2287434240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 793134144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1051709760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12006164480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16754182144.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7828737024.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9447296857833862, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6360543966293335, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2287434240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2367825.432249436, + "step": 175, + "teaching_rms": 1639131.3289531595, + "traffic_rms": 2367825.4322493784 + }, + { + "batch_loss": 3.183842658996582, + "innovation_rms": 1854092.854128519, + "instruction_rms": 0.592428678519642, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14164254720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.831753818533213e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5052045824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.949629545211792, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6788562536239624, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2648185600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 892030080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 988959360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13805940736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17997758464.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5052045824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9541269540786743, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6787090301513672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2648185600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3362835.6307217614, + "step": 176, + "teaching_rms": 1854092.854128519, + "traffic_rms": 3362835.6307217097 + }, + { + "batch_loss": 3.358607053756714, + "innovation_rms": 1943565.1828939416, + "instruction_rms": 0.5859087333442974, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 18306666496.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.801386842791608e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3440061952.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9580680131912231, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5680499076843262, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2962589184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 971111232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 790811200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15665970176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18600294400.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3440061952.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9625649452209473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.567902684211731, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2962589184.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3834982.957368763, + "step": 177, + "teaching_rms": 1943565.1828939416, + "traffic_rms": 3834982.9573687185 + }, + { + "batch_loss": 3.2987051010131836, + "innovation_rms": 2305951.4419510947, + "instruction_rms": 0.7116027289342894, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 22609078272.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.11349319355666e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5315692544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.965643048286438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4545413255691528, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3230385920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1039109120.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 679979072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17664575488.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19986042880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5315692544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.970139503479004, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4543942213058472, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3230385920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4382184.041995555, + "step": 178, + "teaching_rms": 2305951.4419510947, + "traffic_rms": 4382184.041995498 + }, + { + "batch_loss": 3.1356048583984375, + "innovation_rms": 3023313.210343838, + "instruction_rms": 0.9773978770207608, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26276284416.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.417733514539953e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7765560320.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9724409580230713, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0622755289077759, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3472760320.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1131137664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 920284800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20063473664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23988973568.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7765560320.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.976936936378479, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0625121593475342, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3472760320.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5090851.24047124, + "step": 179, + "teaching_rms": 3023313.210343838, + "traffic_rms": 5090851.240471146 + }, + { + "batch_loss": 3.033240795135498, + "innovation_rms": 4184381.306149236, + "instruction_rms": 1.048884237130945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 28109981696.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9891407354802576e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8103837696.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9785393476486206, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0205912590026855, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4283144192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1344366592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2132289280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23411337216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33478633472.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8103837696.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9830348491668701, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0208278894424438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4283144192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5832898.591771189, + "step": 180, + "teaching_rms": 4184381.306149236, + "traffic_rms": 5832898.591771095 + }, + { + "batch_loss": 2.666253089904785, + "innovation_rms": 3152821.682957248, + "instruction_rms": 0.9227439257829219, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29112506368.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1628834518390958e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8643696640.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9840080738067627, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8666769862174988, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5147513856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1537025280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1926586368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 26935027712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35236896768.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8643696640.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.988503098487854, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8669135570526123, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5147513856.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5008168.445878427, + "step": 181, + "teaching_rms": 3152821.682957248, + "traffic_rms": 5008168.445878341 + }, + { + "batch_loss": 2.5911030769348145, + "innovation_rms": 5349940.230447374, + "instruction_rms": 0.08459414992674905, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33714173952.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.942697560285737e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13084850176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9889100790023804, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7674687504768372, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5387266560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1841021952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3039967232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 31492526080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45574979584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13084850176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9934046268463135, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7677052617073059, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5387266560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7727985.097877432, + "step": 182, + "teaching_rms": 5349940.230447374, + "traffic_rms": 7727985.097877431 + }, + { + "batch_loss": 3.487748384475708, + "innovation_rms": 5385400.352739672, + "instruction_rms": 0.9628475250107043, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 38401032192.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.387199521032734e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10867269632.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9933019876480103, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6924802660942078, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6285279744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2200202240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3591801856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 36705738752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52132110336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10867269632.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9977960586547852, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6927167773246765, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6285279744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9045234.522533907, + "step": 183, + "teaching_rms": 5385400.352739672, + "traffic_rms": 9045234.522533856 + }, + { + "batch_loss": 3.193814754486084, + "innovation_rms": 7059924.17957883, + "instruction_rms": 0.8703194653298297, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 44566437888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.323255835123966e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14823329792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.99723482131958, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6298978924751282, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7449515008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2685039104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4848367616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 43161669632.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64559316992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14823329792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0017282962799072, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6300777792930603, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7449515008.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10601266.562844766, + "step": 184, + "teaching_rms": 7059924.17957883, + "traffic_rms": 10601266.56284473 + }, + { + "batch_loss": 2.975266695022583, + "innovation_rms": 6824557.031294944, + "instruction_rms": 0.6273490481900119, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 51652931584.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.708210533094801e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16160788480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.000800609588623, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7555835247039795, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8279827456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3197897728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5128585728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 50277502976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71158341632.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16160788480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.005293607711792, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7555705904960632, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8279827456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10904814.05638237, + "step": 185, + "teaching_rms": 6824557.031294944, + "traffic_rms": 10904814.05638235 + }, + { + "batch_loss": 4.172522068023682, + "innovation_rms": 9994342.180518318, + "instruction_rms": 0.8323262494865011, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 59626323968.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0152753151832438e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30291210240.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.003061532974243, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6753022074699402, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10378665984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3909737984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7118401536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 59377172480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90996686848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30291210240.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.007554054260254, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.675193190574646, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10378665984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13689529.516119162, + "step": 186, + "teaching_rms": 9994342.180518318, + "traffic_rms": 13689529.516119136 + }, + { + "batch_loss": 3.2824931144714355, + "innovation_rms": 8009353.068320669, + "instruction_rms": 0.8209764735341593, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 65386291200.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.501141094432529e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21007974400.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0034351348876953, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7182347774505615, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12479463424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4554801664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6450637312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 68562030592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91848589312.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21007974400.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0133097171783447, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7183926105499268, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12479463424.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17346901.810050577, + "step": 187, + "teaching_rms": 8009353.068320669, + "traffic_rms": 17346901.81005056 + }, + { + "batch_loss": 3.3454737663269043, + "innovation_rms": 8830598.011431491, + "instruction_rms": 1.206234155503292, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 74245554176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2620136961651476e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19391252480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.99970543384552, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5881356596946716, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14156803072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5132328448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5775265280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 77689208832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91271782400.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19391252480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.047192335128784, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5880512595176697, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14156803072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 21383643.090844147, + "step": 188, + "teaching_rms": 8830598.011431491, + "traffic_rms": 21383643.090844113 + }, + { + "batch_loss": 4.9404168128967285, + "innovation_rms": 11936563.514105875, + "instruction_rms": 1.8044368058063383, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 88868880384.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2781549691142417e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36298731520.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9932845830917358, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6639219522476196, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15959463936.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5777604096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6452754432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 87738408960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100491968512.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36298731520.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.075835704803467, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.663790225982666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15959463936.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24724676.02487237, + "step": 189, + "teaching_rms": 11936563.514105875, + "traffic_rms": 24724676.024872303 + }, + { + "batch_loss": 4.24451208114624, + "innovation_rms": 16767581.233488455, + "instruction_rms": 1.5033725386466987, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 110583627776.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2693043603561733e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39255060480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9849169254302979, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6822370290756226, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17622470656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6676951040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8993470464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 100811710464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130732982272.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39255060480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.101811170578003, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.682342529296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17622470656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26015085.11688197, + "step": 190, + "teaching_rms": 16767581.233488455, + "traffic_rms": 26015085.116881926 + }, + { + "batch_loss": 5.0928521156311035, + "innovation_rms": 16257463.557894828, + "instruction_rms": 1.6930290484747048, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 124154732544.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.876201564841273e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46831538176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.978594422340393, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2720658779144287, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19928580096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7730395648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10534448128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 115899957248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150882484224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46831538176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1255149841308594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2718631029129028, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19928580096.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24294626.5249897, + "step": 191, + "teaching_rms": 16257463.557894828, + "traffic_rms": 24294626.524989642 + }, + { + "batch_loss": 5.231077671051025, + "innovation_rms": 24636969.7304788, + "instruction_rms": 1.2106342784870598, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 137142788096.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.071801431316897e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47883071488.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0003137588500977, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9135897159576416, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22653597696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9534157824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18037618688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 136913764352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210138038272.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47883071488.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.147219657897949, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9137012362480164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22653597696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30992277.383593526, + "step": 192, + "teaching_rms": 24636969.7304788, + "traffic_rms": 30992277.383593503 + }, + { + "batch_loss": 4.52946138381958, + "innovation_rms": 24867158.516558565, + "instruction_rms": 1.3983948560794175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 170926931968.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.813108747500113e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42560036864.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0202906131744385, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7927453517913818, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25231923200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11524419584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19902621696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 161333690368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 244199243776.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42560036864.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.167181968688965, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7927974462509155, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25231923200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41128453.82522598, + "step": 193, + "teaching_rms": 24867158.516558565, + "traffic_rms": 41128453.82522595 + }, + { + "batch_loss": 5.088406085968018, + "innovation_rms": 32712636.572059043, + "instruction_rms": 1.2211526370922416, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 202799562752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2043249152137097e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85570158592.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0386130809783936, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2534949779510498, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30249666560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 14051542016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25271228416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 192187318272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 308536246272.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85570158592.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1854896545410156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2535470724105835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30249666560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 46337049.14856257, + "step": 194, + "teaching_rms": 32712636.572059043, + "traffic_rms": 46337049.14856256 + }, + { + "batch_loss": 5.750738620758057, + "innovation_rms": 15885394.598761236, + "instruction_rms": 1.656600509221617, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 188111814656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2619133443688217e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62310109184.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0556488037109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.262545108795166, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 36480675840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15446736896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13951952896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 220511338496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 283240136704.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62310109184.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2025105953216553, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.262605905532837, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 36480675840.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 36045356.922179796, + "step": 195, + "teaching_rms": 15885394.598761236, + "traffic_rms": 36045356.92217976 + }, + { + "batch_loss": 3.81648588180542, + "innovation_rms": 17602028.989314694, + "instruction_rms": 1.6719932959800894, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 173791723520.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3336100874866066e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77986217984.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.072373151779175, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1279652118682861, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41594159104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15657697280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2109601792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 246578167808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 260668342272.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77986217984.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2192203998565674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1276495456695557, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41594159104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 40637875.93319327, + "step": 196, + "teaching_rms": 17602028.989314694, + "traffic_rms": 40637875.93319323 + }, + { + "batch_loss": 4.054687023162842, + "innovation_rms": 45298553.0331185, + "instruction_rms": 1.5260000616572307, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 198643974144.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8592155172578183e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238991409152.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0823888778686523, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2834362983703613, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 56911917056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16864366592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12066689024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 281178898432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 346007437312.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238991409152.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2292213439941406, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.283200740814209, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 56911917056.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 63949722.60212393, + "step": 197, + "teaching_rms": 45298553.0331185, + "traffic_rms": 63949722.602123916 + }, + { + "batch_loss": 4.341590881347656, + "innovation_rms": 39127756.57554863, + "instruction_rms": 1.418881298326907, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 282567737344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2274763002774135e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214356131840.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0910089015960693, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4758448600769043, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 78347534336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17820735488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9563698176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 316693053440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355141386240.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214356131840.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2378265857696533, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.475609302520752, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 78347534336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 90574042.55420594, + "step": 198, + "teaching_rms": 39127756.57554863, + "traffic_rms": 90574042.55420592 + }, + { + "batch_loss": 3.7518668174743652, + "innovation_rms": 38228908.6359847, + "instruction_rms": 1.739536897768505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 401751244800.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.490756239772792e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201354985472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.116183280944824, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4364571571350098, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 98483036160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18114312192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2935772160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 350890229760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 341971861504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201354985472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.254004955291748, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4362215995788574, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 98483036160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 103269442.89339584, + "step": 199, + "teaching_rms": 38228908.6359847, + "traffic_rms": 103269442.89339583 + }, + { + "batch_loss": 3.738774299621582, + "innovation_rms": 41663675.129963316, + "instruction_rms": 2.4203917108476563, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 541450076160.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0061147852496894e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 193856585728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.154536247253418, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5486483573913574, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117868691456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17667911680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4464006144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 383894814720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 330045915136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 193856585728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.27488374710083, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5483322143554688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117868691456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 114074475.80079384, + "step": 200, + "teaching_rms": 41663675.129963316, + "traffic_rms": 114074475.80079381 + }, + { + "batch_loss": 3.021869421005249, + "innovation_rms": 46903814.07606112, + "instruction_rms": 2.5075527093860015, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 690914131968.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5107418624886578e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190847074304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1976661682128906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2359204292297363, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 136953397248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16551118848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11167927296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 416663240704.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 327684358144.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190847074304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2663207054138184, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.235748291015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 136953397248.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 123890177.102542, + "step": 201, + "teaching_rms": 46903814.07606112, + "traffic_rms": 123890177.10254197 + }, + { + "batch_loss": 3.0605738162994385, + "innovation_rms": 54590241.16327603, + "instruction_rms": 2.2843633565306516, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 833750237184.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9346138139453353e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 193077477376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2601377964019775, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.291755199432373, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 156261138432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15019890688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15312281600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 451242786816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 345795461120.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 193077477376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.259310245513916, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2915830612182617, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 156261138432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 132589999.0550902, + "step": 202, + "teaching_rms": 54590241.16327603, + "traffic_rms": 132589999.05509017 + }, + { + "batch_loss": 3.336050271987915, + "innovation_rms": 65142996.12737751, + "instruction_rms": 1.9729906281450607, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 966279823360.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2569479684912878e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200165146624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.294661521911621, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0847113132476807, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 176277651456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13507064832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15128254464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 491026055168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 397832683520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200165146624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.261547327041626, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0845390558242798, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 176277651456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 139162324.08998427, + "step": 203, + "teaching_rms": 65142996.12737751, + "traffic_rms": 139162324.08998427 + }, + { + "batch_loss": 2.9007728099823, + "innovation_rms": 80151796.5864981, + "instruction_rms": 1.4906833251598925, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1074488475648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4290746455294626e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202549379072.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3238792419433594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9746956825256348, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 196532584448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12806137856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7009264640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 541586489344.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 505604210688.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202549379072.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.284952163696289, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9745234847068787, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 196532584448.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 143035147.98475286, + "step": 204, + "teaching_rms": 80151796.5864981, + "traffic_rms": 143035147.98475283 + }, + { + "batch_loss": 3.026761770248413, + "innovation_rms": 98324361.5411293, + "instruction_rms": 1.030194259206255, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1146093174784.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.507554732923991e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 218282393600.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.358031988143921, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.878250241279602, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214853238784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13879653376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10735155200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 609393508352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 678070059008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 218282393600.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.347010374069214, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8780781030654907, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214853238784.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 150314722.43362042, + "step": 205, + "teaching_rms": 98324361.5411293, + "traffic_rms": 150314722.43362042 + }, + { + "batch_loss": 3.2664833068847656, + "innovation_rms": 116456812.16440998, + "instruction_rms": 0.8056251855062885, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1152166264832.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4822547712253193e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192981958656.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3909358978271484, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8498039245605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 233767927808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17704681472.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38250283008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 700890021888.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 914964938752.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192981958656.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.431976079940796, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8496570587158203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 233767927808.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 155999790.21255654, + "step": 206, + "teaching_rms": 116456812.16440998, + "traffic_rms": 155999790.21255654 + }, + { + "batch_loss": 4.131566047668457, + "innovation_rms": 129515008.559384, + "instruction_rms": 0.9844689342052526, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1128488370176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4745745125647868e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 278365339648.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3803634643554688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0039820671081543, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 261585551360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 24703475712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69987934208.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 816683876352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1157938544640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 278365339648.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3315632343292236, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0041289329528809, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 261585551360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 166940534.20616004, + "step": 207, + "teaching_rms": 129515008.559384, + "traffic_rms": 166940534.20616004 + }, + { + "batch_loss": 4.7382893562316895, + "innovation_rms": 156132791.49435118, + "instruction_rms": 0.936033430309457, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1256290254848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.185346656987236e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220496216064.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4170033931732178, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1190948486328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 283635187712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34245644288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95421685760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 959500582912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1428167065600.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220496216064.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.365377426147461, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1190662384033203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 283635187712.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 231662012.50849628, + "step": 208, + "teaching_rms": 156132791.49435118, + "traffic_rms": 231662012.50849628 + }, + { + "batch_loss": 3.552975654602051, + "innovation_rms": 188138321.66036478, + "instruction_rms": 1.060792106346288, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1415911309312.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.240110586419677e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 404839596032.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.468893527984619, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5299534797668457, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 295992197120.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 47207522304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 129618788352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1136624140288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1771236098048.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 404839596032.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4233667850494385, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5299248695373535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 295992197120.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 270174230.855016, + "step": 209, + "teaching_rms": 188138321.66036478, + "traffic_rms": 270174230.855016 + }, + { + "batch_loss": 4.918725490570068, + "innovation_rms": 207525090.98284486, + "instruction_rms": 1.6266448324218774, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1440197640192.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.939728215285055e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 631915085824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.298830509185791, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8674957752227783, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 351615090688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 63238500352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160309772288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1347975380992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2113512013824.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 631915085824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4705870151519775, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8672599792480469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 351615090688.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 272752267.26072323, + "step": 210, + "teaching_rms": 207525090.98284486, + "traffic_rms": 272752267.26072323 + }, + { + "batch_loss": 5.358460903167725, + "innovation_rms": 222597754.72368473, + "instruction_rms": 1.5148088609686745, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1762747482112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.073606496903174e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 399326347264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.342219352722168, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3552141189575195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 390164480000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 79844655104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 166061572096.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1579416289280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2314409082880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 399326347264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.513958692550659, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3549782037734985, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 390164480000.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 376858735.75738055, + "step": 211, + "teaching_rms": 222597754.72368473, + "traffic_rms": 376858735.75738055 + }, + { + "batch_loss": 5.824455261230469, + "innovation_rms": 245852708.54526594, + "instruction_rms": 1.1775537243599628, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2215445528576.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0157448143184972e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 545092599808.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3824431896209717, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9171805381774902, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 415652970496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 96088784896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162441297920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1827608985600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2481927487488.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 545092599808.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5541651248931885, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.916944682598114, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 415652970496.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 442788130.03626543, + "step": 212, + "teaching_rms": 245852708.54526594, + "traffic_rms": 442788130.03626543 + }, + { + "batch_loss": 5.206492900848389, + "innovation_rms": 307374368.5939395, + "instruction_rms": 1.2893712286703851, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2677765308416.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.434744087151142e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 691009945600.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4198477268218994, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9422031044960022, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 421377048576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 114075402240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 179866206208.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2110188290048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2825792520192.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 691009945600.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.591552495956421, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.942101240158081, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 421377048576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 521905602.09523714, + "step": 213, + "teaching_rms": 307374368.5939395, + "traffic_rms": 521905602.09523714 + }, + { + "batch_loss": 5.425127029418945, + "innovation_rms": 441330087.720237, + "instruction_rms": 1.6250140314551373, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3006495719424.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.023919834268122e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1139445989376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4556350708007812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9072262048721313, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 384345636864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 142734950400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286595416064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2489713950720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3795255820288.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1139445989376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6273226737976074, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9074636697769165, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 384345636864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 616335334.4373357, + "step": 214, + "teaching_rms": 441330087.720237, + "traffic_rms": 616335334.4373357 + }, + { + "batch_loss": 6.03110408782959, + "innovation_rms": 226385076.6840312, + "instruction_rms": 1.44924311206578, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2951317553152.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1115859678657936e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 891118747648.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4892520904541016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2239842414855957, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 445358473216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 158949883904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162149400576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2845990977536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3562769743872.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 891118747648.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6609225273132324, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2240798473358154, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 445358473216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 456821305.86181784, + "step": 215, + "teaching_rms": 226385076.6840312, + "traffic_rms": 456821305.86181784 + }, + { + "batch_loss": 5.717033863067627, + "innovation_rms": 229560637.88555336, + "instruction_rms": 1.3175253908144784, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2974206918656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5320597552452365e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 871262715904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5206832885742188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6640596389770508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 524440535040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 160762085376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18122047488.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3175949795328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3299587129344.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 871262715904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6923365592956543, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6641552448272705, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 524440535040.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 504755644.5746933, + "step": 216, + "teaching_rms": 229560637.88555336, + "traffic_rms": 504755644.5746933 + }, + { + "batch_loss": 7.14595890045166, + "innovation_rms": 487893738.03183144, + "instruction_rms": 1.3640218760862832, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3386848575488.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.875957505095009e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2602090627072.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5526130199432373, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3095171451568604, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 759724834816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 167252312064.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64902320128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3567883124736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3919333818368.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2602090627072.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7242491245269775, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3092572689056396, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 759724834816.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 710466064.847457, + "step": 217, + "teaching_rms": 487893738.03183144, + "traffic_rms": 710466064.847457 + }, + { + "batch_loss": 4.9911017417907715, + "innovation_rms": 584674143.3081429, + "instruction_rms": 1.759347317374562, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4687969910784.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.344158379351953e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2268621176832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5824036598205566, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4874950647354126, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 986586939392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 178111004672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108586876928.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4026093273088.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4582101483520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2268621176832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7540225982666016, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4876835346221924, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 986586939392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1035519911.7689819, + "step": 218, + "teaching_rms": 584674143.3081429, + "traffic_rms": 1035519911.7689819 + }, + { + "batch_loss": 7.716791152954102, + "innovation_rms": 560520537.1047194, + "instruction_rms": 2.7319532749223825, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6369966030848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1359530364863443e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2100424474624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.521545886993408, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5968016386032104, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1196629426176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 185463898112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73528983552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4503733010432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4776394752000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2100424474624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.693147659301758, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5968942642211914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1196629426176.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1198687187.5971336, + "step": 219, + "teaching_rms": 560520537.1047194, + "traffic_rms": 1198687187.5971336 + }, + { + "batch_loss": 7.0226545333862305, + "innovation_rms": 595485450.2133509, + "instruction_rms": 2.338924540925135, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8194112356352.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9154407293237234e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2061152550912.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4691197872161865, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3548108339309692, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1402744733696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 186740998144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12770930688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4992831848448.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4890986283008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2061152550912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.640704393386841, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3549035787582397, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1402744733696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1327633475.6317885, + "step": 220, + "teaching_rms": 595485450.2133509, + "traffic_rms": 1327633475.6317885 + }, + { + "batch_loss": 5.864837646484375, + "innovation_rms": 648489569.3373861, + "instruction_rms": 2.237917262822703, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9983195348992.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.636520726984817e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2064939220992.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4257047176361084, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.091233730316162, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1609238708224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 181357756416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53832355840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5494507831296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5016758255616.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2064939220992.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5972721576690674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0913264751434326, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1609238708224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1449264178.3518083, + "step": 221, + "teaching_rms": 648489569.3373861, + "traffic_rms": 1449264178.3518083 + }, + { + "batch_loss": 5.3016581535339355, + "innovation_rms": 758488517.3479275, + "instruction_rms": 2.4690656486878515, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11568592977920.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.167196891081594e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2180349034496.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.390669107437134, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.083780288696289, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1827273572352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 173329317888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80284434432.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6034747293696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5402392526848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2180349034496.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5622193813323975, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0838854312896729, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1827273572352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1581315731.716637, + "step": 222, + "teaching_rms": 758488517.3479275, + "traffic_rms": 1581315731.716637 + }, + { + "batch_loss": 4.847405910491943, + "innovation_rms": 940600854.1860993, + "instruction_rms": 2.3306674410397274, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12792326258688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3983276767961472e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2519111565312.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.363691568374634, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7287805080413818, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2079184781312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 171615731712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17135886336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6672962027520.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6382145241088.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2519111565312.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.535224676132202, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7288856506347656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2079184781312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1738331282.182063, + "step": 223, + "teaching_rms": 940600854.1860993, + "traffic_rms": 1738331282.182063 + }, + { + "batch_loss": 4.7808918952941895, + "innovation_rms": 1303226811.6690555, + "instruction_rms": 1.642452385483869, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13458272681984.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.312033346850732e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3104798932992.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3477394580841064, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8540975451469421, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2389664595968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 201543843840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 299281088512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7582056972288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9090950496256.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3104798932992.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5192553997039795, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.854192852973938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2389664595968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1869848596.660936, + "step": 224, + "teaching_rms": 1303226811.6690555, + "traffic_rms": 1869848596.660936 + }, + { + "batch_loss": 4.077389240264893, + "innovation_rms": 846289945.8163916, + "instruction_rms": 1.515490124414656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13133849559040.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3085040235408296e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1815232118784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3366169929504395, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8060410022735596, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2571187781632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 212123893760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105800433664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8493738754048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9116819914752.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1815232118784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.508115768432617, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8059633374214172, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2571187781632.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1479678324.3055472, + "step": 225, + "teaching_rms": 846289945.8163916, + "traffic_rms": 1479678324.3055472 + }, + { + "batch_loss": 3.469071626663208, + "innovation_rms": 1314225616.185107, + "instruction_rms": 1.2017972565719879, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13409836859392.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2835188310301937e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4436018069504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.334935426712036, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.047567367553711, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2664480899072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 231653916672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195300147200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9570532458496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10767935995904.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4436018069504.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5064170360565186, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.047675609588623, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2664480899072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1921578738.906363, + "step": 226, + "teaching_rms": 1314225616.185107, + "traffic_rms": 1921578738.906363 + }, + { + "batch_loss": 3.945005178451538, + "innovation_rms": 1490178322.1209784, + "instruction_rms": 0.9803442082337497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 16251573239808.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.624310970071435e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2985177120768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.347898483276367, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9918085336685181, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2804757037056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 258458796032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 268048711680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10784138592256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12136064483328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2985177120768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5193629264831543, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9919259548187256, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2804757037056.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2751997945.6873674, + "step": 227, + "teaching_rms": 1490178322.1209784, + "traffic_rms": 2751997945.6873674 + }, + { + "batch_loss": 3.2627031803131104, + "innovation_rms": 1455639730.060778, + "instruction_rms": 0.9741723529718793, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20557170999296.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.405784669391858e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2814574592000.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3702094554901123, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5308635234832764, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2960443834368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 275014418432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 165556125696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12036856610816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12527180185600.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2814574592000.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.541656732559204, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5307259559631348, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2960443834368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3114246558.6030393, + "step": 228, + "teaching_rms": 1455639730.060778, + "traffic_rms": 3114246558.6030393 + }, + { + "batch_loss": 3.193086624145508, + "innovation_rms": 1550990393.1699104, + "instruction_rms": 1.0617513709351747, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 25465077104640.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3454439928448889e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2792428666880.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.398181676864624, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3120445013046265, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3122416058368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 276072562688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10581508096.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13311054708736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12741984124928.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2792428666880.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5696117877960205, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3119069337844849, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3122416058368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3451602518.864972, + "step": 229, + "teaching_rms": 1550990393.1699104, + "traffic_rms": 3451602518.864972 + }, + { + "batch_loss": 3.31718111038208, + "innovation_rms": 1753757019.9883425, + "instruction_rms": 1.438335961668968, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 30188180078592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7784790041322114e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2962858967040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.385268449783325, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1305094957351685, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3322444513280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 264982249472.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110903074816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14641583357952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13305281249280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2962858967040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5566813945770264, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1303718090057373, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3322444513280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3748327779.222374, + "step": 230, + "teaching_rms": 1753757019.9883425, + "traffic_rms": 3748327779.222374 + }, + { + "batch_loss": 2.5810706615448, + "innovation_rms": 2070875654.0279431, + "instruction_rms": 1.303723829295474, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34592358662144.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0816899811461827e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3752753364992.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3576483726501465, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9941970705986023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3609680674816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 252546580480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124356632576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16112540450816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14709574074368.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3752753364992.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5290441513061523, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9940593838691711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3609680674816.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4088549739.9937615, + "step": 231, + "teaching_rms": 2070875654.0279431, + "traffic_rms": 4088549739.9937615 + }, + { + "batch_loss": 3.202209949493408, + "innovation_rms": 2753276290.5968795, + "instruction_rms": 1.3630364078277226, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 38153582804992.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.164982195727322e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6775646453760.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3684561252593994, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8767777681350708, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4119680778240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 278985441280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 264388640768.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17993211838464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18806713876480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6775646453760.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5192694664001465, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8766400814056396, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4119680778240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4573489846.361334, + "step": 232, + "teaching_rms": 2753276290.5968795, + "traffic_rms": 4573489846.361334 + }, + { + "batch_loss": 3.475125789642334, + "innovation_rms": 3363004942.3412833, + "instruction_rms": 1.5016590750411138, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 39394069184512.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0804929719232396e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5783198629888.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.376692056655884, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7453941106796265, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4127102337024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 390934364160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1119489097728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20557066141696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25638534643712.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5783198629888.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.519538640975952, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7452564239501953, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4127102337024.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4392476598.104629, + "step": 233, + "teaching_rms": 3363004942.3412833, + "traffic_rms": 4392476598.104629 + }, + { + "batch_loss": 3.118391513824463, + "innovation_rms": 3962030807.3712354, + "instruction_rms": 1.6886672929334587, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 42000732651520.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0950845309529846e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6241964261376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.37907075881958, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6342218518257141, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4544054427648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 581063475200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1901291372544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23877231050752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33201649090560.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6241964261376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5174736976623535, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.634084165096283, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4544054427648.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5624000342.025578, + "step": 234, + "teaching_rms": 3962030807.3712354, + "traffic_rms": 5624000342.025578 + }, + { + "batch_loss": 3.1699094772338867, + "innovation_rms": 4967882906.878688, + "instruction_rms": 1.573253249510787, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 42699273011200.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.632207232584341e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6199538876416.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.382962703704834, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6588848829269409, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4802149351424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 909611302912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3285478014976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 28308620181504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44313887113216.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6199538876416.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.478372812271118, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6591969132423401, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4802149351424.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6538656565.7894745, + "step": 235, + "teaching_rms": 4967882906.878688, + "traffic_rms": 6538656565.7894745 + }, + { + "batch_loss": 2.5539095401763916, + "innovation_rms": 6014736214.074882, + "instruction_rms": 1.148592995681319, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 44066528034816.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.379943052133106e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10337249132544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3880984783172607, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7623021602630615, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5320332017664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1386436100096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4768248365056.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 34086823395328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57782036332544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10337249132544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4789507389068604, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7626141905784607, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5320332017664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7766310830.083111, + "step": 236, + "teaching_rms": 6014736214.074882, + "traffic_rms": 7766310830.083111 + }, + { + "batch_loss": 2.86086368560791, + "innovation_rms": 6091665216.80286, + "instruction_rms": 0.9954042298233352, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43465626877952.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5608556461974e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10614615310336.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.394059181213379, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9205632209777832, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6068376698880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1932108890112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5456727638016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 40711470710784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66246464765952.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10614615310336.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5059704780578613, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9203174114227295, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6068376698880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8989441497.59849, + "step": 237, + "teaching_rms": 6091665216.80286, + "traffic_rms": 8989441497.59849 + }, + { + "batch_loss": 2.9549288749694824, + "innovation_rms": 8888086500.098581, + "instruction_rms": 1.0401976260446673, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 46615066836992.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.449117496205242e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21670616104960.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3971025943756104, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8288367390632629, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7858465275904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2690718892032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7586098708480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 49376554647552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86650830979072.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21670616104960.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.529541254043579, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.828589677810669, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7858465275904.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11733185792.062904, + "step": 238, + "teaching_rms": 8888086500.098581, + "traffic_rms": 11733185792.062904 + }, + { + "batch_loss": 3.2119646072387695, + "innovation_rms": 3116910612.6897373, + "instruction_rms": 1.4036500700095005, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 38974697504768.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.762650164398216e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19553060913152.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.398188829421997, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5884023904800415, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9813771681792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3106289221632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4155703820288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 57138214862848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77616593764352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19553060913152.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.548218011856079, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5883064270019531, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9813771681792.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8558743733.271369, + "step": 239, + "teaching_rms": 3116910612.6897373, + "traffic_rms": 8558743733.271369 + }, + { + "batch_loss": 3.0140650272369385, + "innovation_rms": 2624853764.283045, + "instruction_rms": 1.6341747243614175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 28260633149440.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.24502387733365e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17782030729216.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4009175300598145, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6207819581031799, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11591975174144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3096033886208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 102553878528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 63908870094848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67706569097216.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17782030729216.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.562734603881836, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6203612089157104, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11591975174144.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10075668536.010647, + "step": 240, + "teaching_rms": 2624853764.283045, + "traffic_rms": 10075668536.010647 + }, + { + "batch_loss": 3.3371634483337402, + "innovation_rms": 2917226694.1963983, + "instruction_rms": 1.9739351559140146, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 18224106176512.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3682888649808845e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15568851697664.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4062659740448, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1894067525863647, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13148859924480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2660704190464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4353297743872.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 69758410031104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58495382585344.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15568851697664.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5732855796813965, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1891758441925049, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13148859924480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11724648630.30694, + "step": 241, + "teaching_rms": 2917226694.1963983, + "traffic_rms": 11724648630.30694 + }, + { + "batch_loss": 3.3384459018707275, + "innovation_rms": 3935517146.70016, + "instruction_rms": 1.8642606890493638, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21680680337408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8833175323002775e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13569796079616.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.409973621368408, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6968897581100464, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14422170927104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873495064576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7872090996736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 74861615513600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51032050630656.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13569796079616.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5804784297943115, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6971354484558105, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14422170927104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13866436209.484833, + "step": 242, + "teaching_rms": 3935517146.70016, + "traffic_rms": 13866436209.484833 + }, + { + "batch_loss": 3.817507743835449, + "innovation_rms": 6039998353.816764, + "instruction_rms": 1.736622532651552, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24659831554048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3253367904810273e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16984634818560.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3940722942352295, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.558039665222168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15328381763584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 894030905344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9794641461248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 79587866312704.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47262495408128.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16984634818560.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5848405361175537, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5578789710998535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15328381763584.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17370676908.951775, + "step": 243, + "teaching_rms": 6039998353.816764, + "traffic_rms": 17370676908.951775 + }, + { + "batch_loss": 4.625663757324219, + "innovation_rms": 10457405808.47075, + "instruction_rms": 1.3219881525110715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26489122717696.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6246219314212317e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33716422311936.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.381760358810425, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2489410638809204, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15838108188672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 77256933376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8167739621376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 85125027069952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55371569823744.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33716422311936.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5969674587249756, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.248780369758606, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15838108188672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22381994876.774254, + "step": 244, + "teaching_rms": 10457405808.47075, + "traffic_rms": 22381994876.774254 + }, + { + "batch_loss": 4.555985927581787, + "innovation_rms": 14661185099.710457, + "instruction_rms": 1.2840761786753843, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 42101098151936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.951038986266906e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31836103245824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.37414813041687, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9100722074508667, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17676541886464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 443725185024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5209821020160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 93200119234560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80750921646080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31836103245824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6074538230895996, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.909911572933197, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17676541886464.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22962947594.1218, + "step": 245, + "teaching_rms": 14661185099.710457, + "traffic_rms": 22962947594.1218 + }, + { + "batch_loss": 4.641958236694336, + "innovation_rms": 16314338691.835798, + "instruction_rms": 1.1803972344983948, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 66879775309824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.646538496539361e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33412893114368.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.371702194213867, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9216773509979248, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21017831407616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 661245132800.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2175199608832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 104081980915712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108818650365952.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33412893114368.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.613609790802002, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9218560457229614, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21017831407616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24533699655.480564, + "step": 246, + "teaching_rms": 16314338691.835798, + "traffic_rms": 24533699655.480564 + }, + { + "batch_loss": 4.207598686218262, + "innovation_rms": 20958447787.57356, + "instruction_rms": 1.1725772225076232, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 77009929633792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.408841488522205e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36331329683456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3738901615142822, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9747524261474609, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22953267822592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 181666414592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4795787116544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 119912064352256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158300800811008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36331329683456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6139633655548096, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9748608469963074, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22953267822592.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26326534310.906525, + "step": 247, + "teaching_rms": 20958447787.57356, + "traffic_rms": 26326534310.906525 + }, + { + "batch_loss": 3.3513238430023193, + "innovation_rms": 22374408649.476772, + "instruction_rms": 1.1927043543899123, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 103241895378944.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.448609841808333e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33391915302912.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.379103899002075, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7926591634750366, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25722351517696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 694892167168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765585752064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 139419981971456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195079142637568.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33391915302912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.609853744506836, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7924171686172485, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25722351517696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35343713928.99611, + "step": 248, + "teaching_rms": 22374408649.476772, + "traffic_rms": 35343713928.99611 + }, + { + "batch_loss": 3.8315584659576416, + "innovation_rms": 26021528093.86996, + "instruction_rms": 1.3425449543232812, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 132947877298176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.872951193176675e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53750894952448.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.389530658721924, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8848695158958435, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 29927592165376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1984264011776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12893718315008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 163139064692736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 237190860767232.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53750894952448.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.615872383117676, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8846741914749146, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 29927592165376.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 39948335652.44792, + "step": 249, + "teaching_rms": 26021528093.86996, + "traffic_rms": 39948335652.44792 + }, + { + "batch_loss": 5.269939422607422, + "innovation_rms": 33581947024.02712, + "instruction_rms": 1.8879461094301389, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 145037438484480.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0897899579847938e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84507004764160.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3483526706695557, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8742738962173462, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35728058744832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4216035475456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22317713326080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 194296368070656.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311572966670336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84507004764160.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.611508369445801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.87446928024292, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35728058744832.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 42479251756.70039, + "step": 250, + "teaching_rms": 33581947024.02712, + "traffic_rms": 42479251756.70039 + }, + { + "batch_loss": 4.97896671295166, + "innovation_rms": 34167288631.721638, + "instruction_rms": 1.647582200453476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 175610961854464.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.556575814758906e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75842008907776.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.319918632507324, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1048455238342285, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43312257957888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6681994461184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24659590381568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 229610260267008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 353138955517952.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75842008907776.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6093852519989014, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1049880981445312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43312257957888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 55351684785.12176, + "step": 251, + "teaching_rms": 34167288631.721638, + "traffic_rms": 55351684785.12176 + }, + { + "batch_loss": 4.158825397491455, + "innovation_rms": 44200283289.92419, + "instruction_rms": 1.6401720056506044, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 207306981638144.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.230690796768658e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127931422605312.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.317570924758911, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4348928928375244, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 52553127559168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9769473540096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30874785546240.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 272448566067200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 428383024447488.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127931422605312.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6318159103393555, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.435035467147827, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 52553127559168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 65721928277.07544, + "step": 252, + "teaching_rms": 44200283289.92419, + "traffic_rms": 65721928277.07544 + }, + { + "batch_loss": 6.091696262359619, + "innovation_rms": 29616514024.663853, + "instruction_rms": 2.095460186980915, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 219990154280960.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3040884139469837e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97330065833984.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3271889686584473, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9106519222259521, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 62286135820288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11885205258240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21157319278592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 314046297407488.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 415977447620608.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97330065833984.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.64140248298645, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9107944965362549, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 62286135820288.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52824297822.686356, + "step": 253, + "teaching_rms": 29616514024.663853, + "traffic_rms": 52824297822.686356 + }, + { + "batch_loss": 6.626300811767578, + "innovation_rms": 58762211978.39095, + "instruction_rms": 1.166524714708832, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 262481272373248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1207669177080876e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198286073921536.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.323005437850952, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4978218078613281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 73799542243328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 15030876635136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31456711671808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 365920140656640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 518738499600384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198286073921536.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6371874809265137, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4979642629623413, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 73799542243328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 82508563963.41739, + "step": 254, + "teaching_rms": 58762211978.39095, + "traffic_rms": 82508563963.41739 + }, + { + "batch_loss": 7.429737091064453, + "innovation_rms": 50501816848.26978, + "instruction_rms": 1.6730923047076562, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 340384379043840.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.827967569922928e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155526352601088.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3120474815368652, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1634938716888428, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 85854752079872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17733496340480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27026199150592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 418630864994304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 527107109158912.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155526352601088.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6261980533599854, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.163636326789856, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 85854752079872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 106722606213.25696, + "step": 255, + "teaching_rms": 50501816848.26978, + "traffic_rms": 106722606213.25696 + }, + { + "batch_loss": 8.355598449707031, + "innovation_rms": 50378685867.913025, + "instruction_rms": 2.0209367305499524, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 439704289804288.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.218963115126639e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 134515213205504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.319915533065796, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.006071925163269, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97683737935872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19519550521344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17860550197248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 469715877101568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 510850121072640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 134515213205504.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6142609119415283, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.006105899810791, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97683737935872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 122630290052.00587, + "step": 256, + "teaching_rms": 50378685867.913025, + "traffic_rms": 122630290052.00587 + }, + { + "batch_loss": 5.642224311828613, + "innovation_rms": 55833508727.10588, + "instruction_rms": 1.7448215453956957, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 552235184422912.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.351955548432636e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116471946543104.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.340750217437744, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8748375773429871, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109122468970496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20457000534016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9374500126720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 519456899989504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 497410094661632.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116471946543104.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6021981239318848, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8750388622283936, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109122468970496.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 140124051448.22815, + "step": 257, + "teaching_rms": 55833508727.10588, + "traffic_rms": 140124051448.22815 + }, + { + "batch_loss": 5.446631908416748, + "innovation_rms": 62232485472.44458, + "instruction_rms": 1.932859899967959, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 675464272150528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5948036376509924e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113017509380096.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3756184577941895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6588162183761597, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 120194584281088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20568703238144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1117037002752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 568522203725824.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 490653037363200.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113017509380096.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5964910984039307, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6586854457855225, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 120194584281088.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 157677362746.15762, + "step": 258, + "teaching_rms": 62232485472.44458, + "traffic_rms": 157677362746.15762 + }, + { + "batch_loss": 3.661562204360962, + "innovation_rms": 69303976980.20546, + "instruction_rms": 2.0643835740612326, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 798481732075520.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.840843857717109e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124096578846720.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4203522205352783, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.410906195640564, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 130916944969728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19933035495424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6356669562880.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 617883122008064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 493609384148992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124096578846720.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.61080002784729, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4107753038406372, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 130916944969728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 172129577323.2865, + "step": 259, + "teaching_rms": 69303976980.20546, + "traffic_rms": 172129577323.2865 + }, + { + "batch_loss": 3.473273754119873, + "innovation_rms": 76824601507.44531, + "instruction_rms": 2.4904398815160387, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 914905947439104.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.069332363829015e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122001825988608.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4701080322265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.202252984046936, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 141393091624960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18681014779904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12520215543808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 668595478593536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 507123800735744.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122001825988608.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6203737258911133, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2024544477462769, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 141393091624960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 184215077975.66595, + "step": 260, + "teaching_rms": 76824601507.44531, + "traffic_rms": 184215077975.66595 + }, + { + "batch_loss": 3.0373620986938477, + "innovation_rms": 84107552672.55267, + "instruction_rms": 2.197618891638174, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1022164165722112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.296566987192389e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125356086394880.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5225634574890137, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1828148365020752, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 152373242626048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16856139169792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18248751906816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 721531084734464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 529356363399168.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125356086394880.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6225600242614746, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.183016300201416, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 152373242626048.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 194223773487.52783, + "step": 261, + "teaching_rms": 84107552672.55267, + "traffic_rms": 194223773487.52783 + }, + { + "batch_loss": 2.905885934829712, + "innovation_rms": 96300167604.67647, + "instruction_rms": 2.5000495725909624, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1134564063838208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.425825350873051e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144349891395584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.556818962097168, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.918532133102417, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 164853310291968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 14837945991168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20181927591936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 779542134259712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 580110495252480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144349891395584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.613913059234619, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.918733537197113, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 164853310291968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 208486707473.4352, + "step": 262, + "teaching_rms": 96300167604.67647, + "traffic_rms": 208486707473.4352 + }, + { + "batch_loss": 3.059303045272827, + "innovation_rms": 112035039852.61487, + "instruction_rms": 3.6953822332260002, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1254877908959232.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.3838454168344e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178214299238400.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5553247928619385, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6180866956710815, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 179736529600512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13206128951296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16318174593024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 846775753637888.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 672336462217216.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178214299238400.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.592906951904297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6180574297904968, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 179736529600512.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 224269031570.81308, + "step": 263, + "teaching_rms": 112035039852.61487, + "traffic_rms": 224269031570.81308 + }, + { + "batch_loss": 2.6933038234710693, + "innovation_rms": 134388350334.98827, + "instruction_rms": 3.5456475160155065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1380324105781248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.976426852570814e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272170517266432.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5596272945404053, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5568531155586243, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 198436095787008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12962465054720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2436638965760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 930188581928960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 834127947366400.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272170517266432.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5684714317321777, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5568238496780396, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 198436095787008.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 242725033333.84155, + "step": 264, + "teaching_rms": 134388350334.98827, + "traffic_rms": 242725033333.84155 + }, + { + "batch_loss": 2.8206863403320312, + "innovation_rms": 171257210961.70883, + "instruction_rms": 2.7670975109724765, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1504626767560704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.986722367082107e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 430303109709824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5635266304016113, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6023237705230713, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 225537322647552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16061147643904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30986827988992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1044565574287360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1143769923584000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 430303109709824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5302488803863525, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6025251150131226, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 225537322647552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 264883524114.05557, + "step": 265, + "teaching_rms": 171257210961.70883, + "traffic_rms": 264883524114.05557 + }, + { + "batch_loss": 2.46866774559021, + "innovation_rms": 215261499272.33505, + "instruction_rms": 2.563263018379393, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1579097205506048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.506262756485851e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 330458139197440.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.555485486984253, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7810490131378174, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 257552948396032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25569559314432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95084108316672.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1209464938037248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1648993234845696.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 330458139197440.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.485705852508545, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7809367179870605, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 257552948396032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 279383794706.7367, + "step": 266, + "teaching_rms": 215261499272.33505, + "traffic_rms": 279383794706.7367 + }, + { + "batch_loss": 2.8190884590148926, + "innovation_rms": 245015504981.55734, + "instruction_rms": 2.868306225394445, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1612839005454336.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.156100415149023e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 485100382846976.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.554853916168213, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6152262687683105, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 269686398779392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 42158237155328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 165886770020352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1428389890097152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2189249252163584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 485100382846976.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5036590099334717, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6153700947761536, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 269686398779392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 309075585191.1526, + "step": 267, + "teaching_rms": 245015504981.55734, + "traffic_rms": 309075585191.1526 + }, + { + "batch_loss": 2.7104287147521973, + "innovation_rms": 261069371483.3258, + "instruction_rms": 2.630471013508244, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1903486153261056.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.96325636134475e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 341861881348096.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5495033264160156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43845969438552856, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 303872593625088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 61365737750528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192074997563392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1685024252362752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2566343354220544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 341861881348096.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5015883445739746, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43837687373161316, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 303872593625088.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 400106587440.26337, + "step": 268, + "teaching_rms": 261069371483.3258, + "traffic_rms": 400106587440.26337 + }, + { + "batch_loss": 2.8868520259857178, + "innovation_rms": 320692775826.5796, + "instruction_rms": 2.890343753925133, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2181285174837248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2353323107346078e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 590473446031360.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5405216217041016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5472670197486877, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 362919938228224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 84878645264384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 235129075138560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1994449534058496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3094253085392896.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 590473446031360.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4785943031311035, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5473370552062988, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 362919938228224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 476798407664.05634, + "step": 269, + "teaching_rms": 320692775826.5796, + "traffic_rms": 476798407664.05634 + }, + { + "batch_loss": 2.8182077407836914, + "innovation_rms": 321470776219.6614, + "instruction_rms": 2.655034778282445, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2178624442597376.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4098517086862198e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 899846584139776.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5197155475616455, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5669795870780945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 374984702689280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 109714125881344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 248354772615168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2341445713264640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3469960718319616.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 899846584139776.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.436824321746826, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5670547485351562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 374984702689280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 445773349057.20764, + "step": 270, + "teaching_rms": 321470776219.6614, + "traffic_rms": 445773349057.20764 + }, + { + "batch_loss": 2.7423083782196045, + "innovation_rms": 392973021002.6427, + "instruction_rms": 3.3067094676221385, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2784225971404800.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0077335285630072e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 689826105917440.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5026533603668213, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5839838981628418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 406988349505536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 135375725002752.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256615957659648.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2730962202918912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3895164359671808.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 689826105917440.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.434230089187622, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5840513110160828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 406988349505536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 648418237907.8325, + "step": 271, + "teaching_rms": 392973021002.6427, + "traffic_rms": 648418237907.8325 + }, + { + "batch_loss": 2.633946418762207, + "innovation_rms": 435109454885.83856, + "instruction_rms": 3.0802899493364992, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3517024164642816.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.770033793379208e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 781231935455232.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4905688762664795, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4553600549697876, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 473170205212672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 160617147138048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 252414238130176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3158145287323648.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4271831917789184.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 781231935455232.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.459500551223755, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4551309645175934, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 473170205212672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 748470318279.1506, + "step": 272, + "teaching_rms": 435109454885.83856, + "traffic_rms": 748470318279.1506 + }, + { + "batch_loss": 2.599663734436035, + "innovation_rms": 594900980205.4583, + "instruction_rms": 2.0843406400410696, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4074653190455296.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.916359318434007e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1427205284429824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4785211086273193, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47080233693122864, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 557058399141888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 193938392612864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 333212488302592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3685781482766336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5276361954426880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1427205284429824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4789538383483887, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4708421230316162, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 557058399141888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 884224641750.5643, + "step": 273, + "teaching_rms": 594900980205.4583, + "traffic_rms": 884224641750.5643 + }, + { + "batch_loss": 2.895754098892212, + "innovation_rms": 533758261393.82294, + "instruction_rms": 1.5245285307053151, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3960510676467712.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4179087910691676e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1556746061479936.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.45917010307312, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5334426760673523, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 643541390852096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 222682595459072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287442062016512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4254592455933952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5688110805417984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1556746061479936.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.478808879852295, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5334137678146362, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 643541390852096.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 812884845398.0223, + "step": 274, + "teaching_rms": 533758261393.82294, + "traffic_rms": 812884845398.0223 + }, + { + "batch_loss": 2.3637542724609375, + "innovation_rms": 924097357120.2264, + "instruction_rms": 1.5508676393630367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4415469851246592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.028938918255016e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1903438505967616.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.442807197570801, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5105206370353699, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 760330879762432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 277377192034304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 546946032861184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5044441773506560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7898494786338816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1903438505967616.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.477027416229248, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5104377865791321, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 760330879762432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1222914735697.6338, + "step": 275, + "teaching_rms": 924097357120.2264, + "traffic_rms": 1222914735697.6338 + }, + { + "batch_loss": 2.6887388229370117, + "innovation_rms": 1052835812140.0055, + "instruction_rms": 1.1983358524287249, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4954456839946240.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.451093124476283e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2776275986939904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4282398223876953, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44247275590896606, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 987940725456896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 357273499074560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 798963103956992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6061507209068544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0170656503103488e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2776275986939904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.486109733581543, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44250020384788513, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 987940725456896.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1371106689537.711, + "step": 276, + "teaching_rms": 1052835812140.0055, + "traffic_rms": 1371106689537.711 + }, + { + "batch_loss": 2.64987850189209, + "innovation_rms": 1051405783794.9537, + "instruction_rms": 1.1295926717954636, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5246307383902208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.385466828656519e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2727401138159616.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4124605655670166, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5621763467788696, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1163485836738560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 446810950729728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 895374516551680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7206770439094272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1452631226515456e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2727401138159616.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.493664503097534, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5621195435523987, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1163485836738560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1665498011030.9807, + "step": 277, + "teaching_rms": 1051405783794.9537, + "traffic_rms": 1665498011030.9807 + }, + { + "batch_loss": 2.5996406078338623, + "innovation_rms": 1454083280428.344, + "instruction_rms": 1.0314111680549072, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6475128665800704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0233311499339661e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3213578819600384.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.398977756500244, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7118824124336243, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1415430497370112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 562256383311872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1154454460039168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8624466665930752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4176963342106624e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3213578819600384.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.497809886932373, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7120504975318909, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1415430497370112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2064826709015.2087, + "step": 278, + "teaching_rms": 1454083280428.344, + "traffic_rms": 2064826709015.2087 + }, + { + "batch_loss": 3.0978729724884033, + "innovation_rms": 1152173335634.5598, + "instruction_rms": 1.26700296295527, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6816534005547008.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.236882697587866e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4602675932954624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.406496286392212, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.191275954246521, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1589436668182528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 657817258164224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 955608480088064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0081332994506752e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4568660064534528e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4602675932954624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4987728595733643, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1913166046142578, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1589436668182528.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1803457383526.771, + "step": 279, + "teaching_rms": 1152173335634.5598, + "traffic_rms": 1803457383526.771 + }, + { + "batch_loss": 3.8058114051818848, + "innovation_rms": 1715914864377.4136, + "instruction_rms": 1.5327025451026466, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9743046337888256.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5631840921695353e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3094122357325824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4282915592193604, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0749881267547607, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1898848930758656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 762488362631168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1046711178887168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1728370274402304e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6470370651471872e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3094122357325824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.499631643295288, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0749599933624268, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1898848930758656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2934296302609.815, + "step": 280, + "teaching_rms": 1715914864377.4136, + "traffic_rms": 2934296302609.815 + }, + { + "batch_loss": 4.068282604217529, + "innovation_rms": 1970952388213.9148, + "instruction_rms": 2.1725431396277215, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3238142547001344e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.442327747631302e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4332899910287360.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4586100578308105, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7519958019256592, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2332139056005120.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 876495719366656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1140073433137152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3583743532924928e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8553736880193536e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4332899910287360.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5019984245300293, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7522295713424683, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2332139056005120.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3483290201097.091, + "step": 281, + "teaching_rms": 1970952388213.9148, + "traffic_rms": 3483290201097.091 + }, + { + "batch_loss": 3.69356369972229, + "innovation_rms": 2933921679396.9565, + "instruction_rms": 3.3490699140878517, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5531832250466304e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.8285722630919e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6844932329308160.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3239917755126953, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.724652647972107, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2882694270681088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1071638497984512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1951428054614016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.612829475995648e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5445516565282816e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6844932329308160.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.502347707748413, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7243497371673584, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2882694270681088.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3932904686313.7485, + "step": 282, + "teaching_rms": 2933921679396.9565, + "traffic_rms": 3932904686313.7485 + }, + { + "batch_loss": 5.8331708908081055, + "innovation_rms": 1325013678385.2354, + "instruction_rms": 4.012837766143188, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5269975610621952e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.967484126179883e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4402849358282752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.264617443084717, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6168303489685059, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3216830378278912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1179170654650368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1075322170638336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.847259206307021e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.344298162107187e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4402849358282752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4994070529937744, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6169259548187256, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3216830378278912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2960033243835.289, + "step": 283, + "teaching_rms": 1325013678385.2354, + "traffic_rms": 2960033243835.289 + }, + { + "batch_loss": 7.504837512969971, + "innovation_rms": 1794370778997.6638, + "instruction_rms": 4.152443101326175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5936804716806144e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.46106290910116e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5670000002072576.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.242121696472168, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.549020767211914, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3380119397728256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1222777189171200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 436065076772864.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.073592171646157e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.263328794397901e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5670000002072576.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4927241802215576, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.548984408378601, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3380119397728256.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3442432043338.061, + "step": 284, + "teaching_rms": 1794370778997.6638, + "traffic_rms": 3442432043338.061 + }, + { + "batch_loss": 6.359536647796631, + "innovation_rms": 4134529752046.8037, + "instruction_rms": 3.87344741958381, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.101034650435584e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3181856337694323e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2745979023327232e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2252871990203857, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6441797018051147, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4004365782220800.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1359015430848512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1362382819426304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.385946459976499e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1235424538066944e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2745979023327232e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5157461166381836, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6440629959106445, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4004365782220800.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5901644666540.01, + "step": 285, + "teaching_rms": 4134529752046.8037, + "traffic_rms": 5901644666540.01 + }, + { + "batch_loss": 8.541296005249023, + "innovation_rms": 3898856185544.41, + "instruction_rms": 3.6988169338624584, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.596841165370163e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2857243645550183e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8193770566713344.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2120397090911865, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1012301445007324, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4444539800518656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1535891445121024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1768759874289664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7378756507140096e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.51929083363328e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8193770566713344.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5217885971069336, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1013731956481934, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4444539800518656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7292007391244.225, + "step": 286, + "teaching_rms": 3898856185544.41, + "traffic_rms": 7292007391244.225 + }, + { + "batch_loss": 9.148499488830566, + "innovation_rms": 4404672532042.585, + "instruction_rms": 2.484277333065391, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.136802735836365e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.536903398848031e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9847408674471936.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.20345139503479, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.446024179458618, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4900806792839168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1760857537118208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2249661322625024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1322376150974464e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.943619643834368e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9847408674471936.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.510789155960083, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.446167230606079, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4900806792839168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8390995364066.092, + "step": 287, + "teaching_rms": 4404672532042.585, + "traffic_rms": 8390995364066.092 + }, + { + "batch_loss": 8.919160842895508, + "innovation_rms": 6504532675220.747, + "instruction_rms": 1.9813968887954687, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.751261802541875e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.770762878278917e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.340409364086784e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.198662281036377, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8820202350616455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6004969870196736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2136884339802112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3760268295274496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.669581402996736e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.373437019999437e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.340409364086784e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.481365203857422, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8821632862091064, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6004969870196736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9037902641816.133, + "step": 288, + "teaching_rms": 6504532675220.747, + "traffic_rms": 9037902641816.133 + }, + { + "batch_loss": 7.122723579406738, + "innovation_rms": 4068224534637.469, + "instruction_rms": 1.8756921015685601, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.126373938253005e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0291120667837435e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9460555601412096.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1968603134155273, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5155644416809082, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6904242069569536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2456200796962816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3193165376913408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.202716552442675e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.331349346975744e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9460555601412096.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.425567150115967, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5157074928283691, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6904242069569536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7056456745540.297, + "step": 289, + "teaching_rms": 4068224534637.469, + "traffic_rms": 7056456745540.297 + }, + { + "batch_loss": 7.7410125732421875, + "innovation_rms": 4873390438377.807, + "instruction_rms": 2.9744154477299034, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.650921730087322e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.783258175303676e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0243366277021696e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2044050693511963, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5637519359588623, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7611435278401536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2733395570327552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2771948538953728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.742051057683661e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.393343763919667e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0243366277021696e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3488426208496094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.563657283782959, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7611435278401536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8725100048580.206, + "step": 290, + "teaching_rms": 4873390438377.807, + "traffic_rms": 8725100048580.206 + }, + { + "batch_loss": 8.13406753540039, + "innovation_rms": 9534462662799.725, + "instruction_rms": 3.796153437294842, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.55541091886039e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.307782334005631e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4285684269318144e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.219452381134033, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5726685523986816, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9168287371886592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3227629268238336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4942336979107840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.480696956269363e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.386460703843942e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4285684269318144e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.276510000228882, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.572673797607422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9168287371886592.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13366863720878.73, + "step": 291, + "teaching_rms": 9534462662799.725, + "traffic_rms": 13366863720878.73 + }, + { + "batch_loss": 7.107175350189209, + "innovation_rms": 7145439951179.104, + "instruction_rms": 4.878034986152451, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.728239973354701e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.10987552907757e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5426139547762688e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.160581350326538, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9802539348602295, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0710901111914496e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3634063634071552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4064343658332160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.238862622221926e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.581654941538714e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5426139547762688e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2723376750946045, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.980259418487549, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0710901111914496e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14147338932279.238, + "step": 292, + "teaching_rms": 7145439951179.104, + "traffic_rms": 14147338932279.238 + }, + { + "batch_loss": 4.3914031982421875, + "innovation_rms": 8622523860282.973, + "instruction_rms": 3.9426137594041717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.029595065188352e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6060698278452868e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.935244215602381e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.153576612472534, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4211466312408447, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1966617479020544e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4019499904794624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3854363244101632.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.038521537724416e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.996587437037978e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.935244215602381e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.265321731567383, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.421151876449585, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1966617479020544e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17701099216666.18, + "step": 293, + "teaching_rms": 8622523860282.973, + "traffic_rms": 17701099216666.18 + }, + { + "batch_loss": 4.1768975257873535, + "innovation_rms": 13653612246640.762, + "instruction_rms": 2.5921120669455378, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.876082868112589e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.378213572883865e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4111334639468544e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1500535011291504, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.010742664337158, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3308774357925888e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4669131425054720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6496317350084608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.094719708810445e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0561977415892992e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4111334639468544e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2617874145507812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0107481479644775, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3308774357925888e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20993669334356.605, + "step": 294, + "teaching_rms": 13653612246640.762, + "traffic_rms": 20993669334356.605 + }, + { + "batch_loss": 3.5078752040863037, + "innovation_rms": 12069228248765.887, + "instruction_rms": 1.777379337171332, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.59835635136594e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.800748969260305e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.222867625489203e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2195053100585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6943225860595703, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5177828161028096e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5375377593597952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7062461685432320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.25657708185518e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.161857373044736e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.222867625489203e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2601797580718994, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6943281888961792, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5177828161028096e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19853480432765.457, + "step": 295, + "teaching_rms": 12069228248765.887, + "traffic_rms": 19853480432765.457 + }, + { + "batch_loss": 3.3462252616882324, + "innovation_rms": 17396040662464.398, + "instruction_rms": 1.4000164490990457, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.349869971970458e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4641591794032814e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.209963263269274e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2964396476745605, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.521535873413086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.810371450188595e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6402926507458560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0275488064864256e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0738864784985293e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4822873595327283e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.209963263269274e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2560927867889404, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5215413570404053, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.810371450188595e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24432298353941.574, + "step": 296, + "teaching_rms": 17396040662464.398, + "traffic_rms": 24432298353941.574 + }, + { + "batch_loss": 3.659221887588501, + "innovation_rms": 18094178817346.508, + "instruction_rms": 1.698861143154008, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.092991226478592e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.268008652367502e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.092863072914637e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.369978189468384, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3692469596862793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.035994886930432e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7656534045622272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2536076455378944e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2428761758275994e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6898973168880845e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.092863072914637e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.300248622894287, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.369252324104309, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.035994886930432e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 32045779362862.887, + "step": 297, + "teaching_rms": 18094178817346.508, + "traffic_rms": 32045779362862.887 + }, + { + "batch_loss": 3.812203884124756, + "innovation_rms": 20800575371405.074, + "instruction_rms": 1.585817909674392, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.028437478169641e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.73426189022276e+33, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.05471436891095e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4304144382476807, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.232173204421997, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.191365114114867e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9125803702878208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4692698720043008e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4347667246782874e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.91890548850688e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.05471436891095e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.360691785812378, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2321785688400269, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.191365114114867e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 37683933447103.375, + "step": 298, + "teaching_rms": 20800575371405.074, + "traffic_rms": 37683933447103.375 + }, + { + "batch_loss": 4.287206172943115, + "innovation_rms": 29079114473854.957, + "instruction_rms": 1.4663416012390826, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2185174406070272e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0368248575760951e+34, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.01657229485015e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4675121307373047, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1087944507598877, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3138045385506816e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1155336811511808e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0295335381303296e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6856724434583552e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.509057703196754e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.01657229485015e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.397796392440796, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1087998151779175, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3138045385506816e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41443024493624.09, + "step": 299, + "teaching_rms": 29079114473854.957, + "traffic_rms": 41443024493624.09 + }, + { + "batch_loss": 3.404052972793579, + "innovation_rms": 28813940244934.703, + "instruction_rms": 1.2802243311221682, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3732934590660608e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1692273108468177e+34, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.974747340845875e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.511240243911743, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1189416646957397, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9749395918422016e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.351157519745024e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.356238171190067e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9777805566030643e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.921080444252324e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.974747340845875e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4415314197540283, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1187609434127808, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9749395918422016e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 39440055189320.83, + "step": 300, + "teaching_rms": 28813940244934.703, + "traffic_rms": 39440055189320.83 + }, + { + "batch_loss": 3.7177019119262695, + "innovation_rms": 43215332344378.836, + "instruction_rms": 1.1601879782573492, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6976916517381734e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.061835000069689e+33, + "nonfinite_indices": [ + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.935529806974157e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.562462091445923, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.09159517288208, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.968492658438963e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6938600469889024e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.427025272438784e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3718965482094592e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.941159572466565e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.935529806974157e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.492760181427002, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.091774344444275, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.968492658438963e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 56381130198530.76, + "step": 301, + "teaching_rms": 43215332344378.836, + "traffic_rms": 56381130198530.76 + }, + { + "batch_loss": 3.1790716648101807, + "innovation_rms": 41990742007180.27, + "instruction_rms": 1.3106395675635756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.803576854875996e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.8762846441053e+33, + "nonfinite_indices": [ + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2453761044919091e+17, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6102960109710693, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0921401977539062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.347955166524211e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0712913576656896e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.774312462522778e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8238394640931226e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5194291588366336e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2453761044919091e+17, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5406012535095215, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0923192501068115, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.347955166524211e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 62129347275250.11, + "step": 302, + "teaching_rms": 41990742007180.27, + "traffic_rms": 62129347275250.11 + }, + { + "batch_loss": 3.702026128768921, + "innovation_rms": 1311758775136.3054, + "instruction_rms": 1.6657931486666855, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5021601575180698e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27209.826171875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1208016088635802e+17, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6243250370025635, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9091581106185913, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.80383502123008e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3738728774107136e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0258145532051456e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.230588157107896e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.06748624295297e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1208016088635802e+17, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5546371936798096, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9090908169746399, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.80383502123008e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6587632519136.7295, + "step": 303, + "teaching_rms": 1311758775136.3054, + "traffic_rms": 6587632519136.7295 + }, + { + "batch_loss": 2.856395721435547, + "innovation_rms": 1503385135554.5295, + "instruction_rms": 1.6017039338180512, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6856633067411866e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 29012.794921875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0086734130629837e+17, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6319046020507812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9451942443847656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.81250852019241e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6036688748806144e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.297959974699008e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5966619636413235e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6607373781395046e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0086734130629837e+17, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5622236728668213, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.945127010345459, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.81250852019241e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7549977400305.455, + "step": 304, + "teaching_rms": 1503385135554.5295, + "traffic_rms": 7549977400305.455 + }, + { + "batch_loss": 2.756110668182373, + "innovation_rms": 1648915933687.7974, + "instruction_rms": 1.5828848259719943, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.2234883729824154e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 30503.814453125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.0774786635989e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6454672813415527, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7811537384986877, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.720256300652954e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.763841304998707e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6017234421874688e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.926128217722716e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2946635716060774e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.0774786635989e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.575793504714966, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7810865640640259, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.720256300652954e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8280830993833.627, + "step": 305, + "teaching_rms": 1648915933687.7974, + "traffic_rms": 8280830993833.627 + }, + { + "batch_loss": 3.0536513328552246, + "innovation_rms": 1750353998181.8367, + "instruction_rms": 1.6109747668347518, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.999356332247613e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 31552.802734375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.169058720756531e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.671168088912964, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6108850240707397, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.53716251632599e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8584828421013504e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9464160152715264.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2226479494751846e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.965196973927301e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.169058720756531e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.601501226425171, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6107044816017151, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.53716251632599e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8790251152231.012, + "step": 306, + "teaching_rms": 1750353998181.8367, + "traffic_rms": 8790251152231.012 + }, + { + "batch_loss": 3.494131088256836, + "innovation_rms": 1810290946448.5408, + "instruction_rms": 1.7437503337564042, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.918584115100385e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 32082.279296875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.351399167819776e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.708743095397949, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7781113386154175, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.272302862604698e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.892451308948685e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3396845610991616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.489515570613453e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6686772421748326e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.351399167819776e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6390833854675293, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7782567143440247, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.272302862604698e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9091253856357.625, + "step": 307, + "teaching_rms": 1810290946448.5408, + "traffic_rms": 9091253856357.625 + }, + { + "batch_loss": 3.6805436611175537, + "innovation_rms": 1831803387554.9172, + "instruction_rms": 1.5092261582339774, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0855722942369628e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 32033.927734375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.615432040336589e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.619041919708252, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9646860361099243, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.933846367286067e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8712053942255616e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2124590398570496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.729696670156063e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4018094492378726e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.615432040336589e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.549389123916626, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9643762111663818, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.933846367286067e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9199288314302.055, + "step": 308, + "teaching_rms": 1831803387554.9172, + "traffic_rms": 9199288314302.055 + }, + { + "batch_loss": 4.06059455871582, + "innovation_rms": 1818348150545.9204, + "instruction_rms": 1.077047202586497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2703744977818092e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 31609.888671875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.952995182457651e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5530970096588135, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7019176483154297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.529146014380851e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.800647241990144e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7055821129121792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.945859625384673e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1616285214939546e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.952995182457651e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4834511280059814, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7016735672950745, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.529146014380851e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9131716878165.963, + "step": 309, + "teaching_rms": 1818348150545.9204, + "traffic_rms": 9131716878165.963 + }, + { + "batch_loss": 4.295765399932861, + "innovation_rms": 1773663403463.889, + "instruction_rms": 1.232870998547259, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4502573488650322e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 30935.669921875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.356742482119885e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.512589693069458, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9892271757125854, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0064820348492186e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6869719130701824e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1367526449545216e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.140406182011208e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.945465566265344e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.356742482119885e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.44295072555542, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9891583919525146, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0064820348492186e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8907310352431.25, + "step": 310, + "teaching_rms": 1773663403463.889, + "traffic_rms": 8907310352431.25 + }, + { + "batch_loss": 3.8537802696228027, + "innovation_rms": 1701672259398.7644, + "instruction_rms": 1.3666382753509014, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6352456447528796e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 30079.271484375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.820061837171098e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.493788003921509, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4506874084472656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0546826489259622e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5365277986258944e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5044413591912448e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.3154982547737805e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7509190096388096e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.820061837171098e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4340054988861084, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4506187438964844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0546826489259622e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8545772257801.934, + "step": 311, + "teaching_rms": 1701672259398.7644, + "traffic_rms": 8545772257801.934 + }, + { + "batch_loss": 3.379138708114624, + "innovation_rms": 1606395407773.369, + "instruction_rms": 1.3910418105135907, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.823407543515349e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 29143.642578125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.337000852435763e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4922752380371094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2139723300933838, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0980526273855488e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3556869121376256e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8084095091277824e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.473080948461404e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5758271086749286e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.337000852435763e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4620237350463867, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2141317129135132, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0980526273855488e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8067292225774.687, + "step": 312, + "teaching_rms": 1606395407773.369, + "traffic_rms": 8067292225774.687 + }, + { + "batch_loss": 3.3933894634246826, + "innovation_rms": 1491867839259.123, + "instruction_rms": 1.7961017341901608, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0142914207513313e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 28232.228515625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.902202415205581e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.523283004760742, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9909117221832275, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1370746386527027e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.150728495792128e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0495837339582464e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.614905235341312e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.418244329087959e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.902202415205581e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.487215518951416, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9910710453987122, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1370746386527027e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7492137671135.153, + "step": 313, + "teaching_rms": 1491867839259.123, + "traffic_rms": 7492137671135.153 + }, + { + "batch_loss": 3.9649484157562256, + "innovation_rms": 1362066644465.2925, + "instruction_rms": 2.383113676544935, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2006645515343626e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27411.04296875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.510844995194061e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.524592161178589, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7418479919433594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1721830757197414e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.927736156015821e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2299227535179776e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.742547196612444e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2764198704093594e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.510844995194061e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5098633766174316, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7420918941497803, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1721830757197414e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6840278211907.43, + "step": 314, + "teaching_rms": 1362066644465.2925, + "traffic_rms": 6840278211907.43 + }, + { + "batch_loss": 3.8712611198425293, + "innovation_rms": 1220844638654.8481, + "instruction_rms": 2.9671257874600694, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.371936852182565e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26769.486328125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1585883132002304e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5357213020324707, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2355642318725586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2037689524092928e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6925082060324864e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.352279285084979e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.857424858677248e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1487778232388813e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1585883132002304e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5302212238311768, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2351467609405518, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2037689524092928e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6131063084286.562, + "step": 315, + "teaching_rms": 1220844638654.8481, + "traffic_rms": 6131063084286.562 + }, + { + "batch_loss": 3.7337825298309326, + "innovation_rms": 1071873568531.4731, + "instruction_rms": 2.521951837806514, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.52508727705967e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26334.7890625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.841525666971648e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5540175437927246, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.133857011795044, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.232184198341591e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4504822891872256e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4202595979493376e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.960814685816095e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0338999893753856e+17, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.841525666971648e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.548517942428589, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1334396600723267, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.232184198341591e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5382932837750.842, + "step": 316, + "teaching_rms": 1071873568531.4731, + "traffic_rms": 5382932837750.842 + }, + { + "batch_loss": 3.2854020595550537, + "innovation_rms": 918597433123.386, + "instruction_rms": 1.9248413458376215, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6690771208099594e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26064.169921875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5561407667830784e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5704586505889893, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.809615969657898, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2577456103043891e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2066739609141248e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4380837122277376e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.05386566768001e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.30509990437847e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5561407667830784e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5649595260620117, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8091986775398254, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2577456103043891e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4613182244754.958, + "step": 317, + "teaching_rms": 918597433123.386, + "traffic_rms": 4613182244754.958 + }, + { + "batch_loss": 3.318406820297241, + "innovation_rms": 764192437566.5042, + "instruction_rms": 0.7449933209783343, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8244609286809846e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 25986.5, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.299268865982464e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5852298736572266, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.712548017501831, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2807382882267955e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9656291219734528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4104479599099904e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.137611345199104e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.374590085739315e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.299268865982464e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5797312259674072, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7128598690032959, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2807382882267955e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3837762713133.6562, + "step": 318, + "teaching_rms": 764192437566.5042, + "traffic_rms": 3837762713133.6562 + }, + { + "batch_loss": 4.367030143737793, + "innovation_rms": 611537579612.6832, + "instruction_rms": 0.6789742133770322, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9905556290739896e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26088.525390625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.068061112762368e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5984981060028076, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5056228637695312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3014188993544192e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7313898197745664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3423929146146816e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.212982867283149e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.537130733568e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.068061112762368e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5930001735687256, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.505850911140442, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3014188993544192e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3071132353098.192, + "step": 319, + "teaching_rms": 611537579612.6832, + "traffic_rms": 3071132353098.192 + }, + { + "batch_loss": 4.783128261566162, + "innovation_rms": 463192685126.2008, + "instruction_rms": 0.6762415101069177, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1602787178386555e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26352.3359375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859953626395443e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6104135513305664, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.354646921157837, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.320018427028439e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5074718071717888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.239180126027776e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.280817237158789e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.7834176602112e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859953626395443e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6049160957336426, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3544948101043701, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.320018427028439e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2326146722831.202, + "step": 320, + "teaching_rms": 463192685126.2008, + "traffic_rms": 2326146722831.202 + }, + { + "batch_loss": 4.381250381469727, + "innovation_rms": 321384340252.371, + "instruction_rms": 0.6883420379311149, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3173703421455237e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26627.21484375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6726381840826368e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6211113929748535, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1022981405258179, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3367448335653274e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2968543687606272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1061742230503424e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.341868307485819e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.105075679441715e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6726381840826368e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.615614414215088, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.102146029472351, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3367448335653274e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1613987223858.9526, + "step": 321, + "teaching_rms": 321384340252.371, + "traffic_rms": 1613987223858.9526 + }, + { + "batch_loss": 4.2254958152771, + "innovation_rms": 187999306399.16638, + "instruction_rms": 0.7657056286929292, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.456688086111486e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26914.337890625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5040376322654208e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6307132244110107, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8095412254333496, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.351785207740498e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1019806161567744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.948737472351437e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.396813652304855e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.494567982648525e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5040376322654208e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6252167224884033, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8098962306976318, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.351785207740498e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 944129630938.7628, + "step": 322, + "teaching_rms": 187999306399.16638, + "traffic_rms": 944129630938.7628 + }, + { + "batch_loss": 4.357901573181152, + "innovation_rms": 64584815070.50162, + "instruction_rms": 1.086741100641491, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.575120606707843e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27133.603515625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3522820495572992e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6393284797668457, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8275766372680664, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3653079981712998e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 752327141097472.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7721332489781248e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.446264874958848e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.945111098484326e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3522820495572992e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6338326930999756, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8279987573623657, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3653079981712998e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 324343952313.9717, + "step": 323, + "teaching_rms": 64584815070.50162, + "traffic_rms": 324343952313.9717 + }, + { + "batch_loss": 3.6836187839508057, + "innovation_rms": 1.288813169265937, + "instruction_rms": 1.2888124546817934, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.676929885881762e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27224.05859375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2156885603975168e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6470558643341064, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9609711170196533, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3774649031026278e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2347247140339712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5949198918680576e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.490771044066918e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.450599730937856e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2156885603975168e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6415605545043945, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9611742496490479, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3774649031026278e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.2888220749446628, + "step": 324, + "teaching_rms": 1.288813169265937, + "traffic_rms": 0.004987924046494246 + }, + { + "batch_loss": 3.0870285034179688, + "innovation_rms": 2.1805320104092933, + "instruction_rms": 2.1805310077208606, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7731148882016993e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27163.345703125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.092742222446592e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6539840698242188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9446864128112793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3883923306958029e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3782675139657728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4354278919438336e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.530826252666798e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.00553975784407e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.092742222446592e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.648489236831665, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9448895454406738, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3883923306958029e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.1805368464732537, + "step": 325, + "teaching_rms": 2.1805320104092933, + "traffic_rms": 0.005091514919230705 + }, + { + "batch_loss": 3.809065103530884, + "innovation_rms": 2.5611753161662834, + "instruction_rms": 2.561174966199351, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8755357707185684e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27147.275390625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9820795983364096.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6601929664611816, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6160850524902344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3982131170154906e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5074560231669760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2918850383249408e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.566875802967736e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.604985524361626e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9820795983364096.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.654698610305786, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6160545349121094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3982131170154906e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.561179960077052, + "step": 326, + "teaching_rms": 2.5611753161662834, + "traffic_rms": 0.0050796373955823695 + }, + { + "batch_loss": 5.620018005371094, + "innovation_rms": 2.459811170569769, + "instruction_rms": 2.459809511600436, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9809311070664786e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27213.7421875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824734011621376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6657543182373047, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2677507400512695, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.40703781451989e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6237256733949952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.162696502280192e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.599320329519104e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.244486800126771e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824734011621376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6602604389190674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2677910327911377, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.40703781451989e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.4598145501571533, + "step": 327, + "teaching_rms": 2.459811170569769, + "traffic_rms": 0.005090772453762295 + }, + { + "batch_loss": 4.8975605964660645, + "innovation_rms": 1.9774144803161828, + "instruction_rms": 1.977412691895821, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0746441323692687e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27228.337890625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7928190190223360.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6707329750061035, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6884660720825195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4149659805509222e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7283683639689216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0464267983650816e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.628520609573765e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9200380986392576e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7928190190223360.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6652395725250244, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6885063648223877, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4149659805509222e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.9774190306189625, + "step": 328, + "teaching_rms": 1.9774144803161828, + "traffic_rms": 0.005122406138830753 + }, + { + "batch_loss": 5.090121269226074, + "innovation_rms": 1.2132298119507232, + "instruction_rms": 1.213231835267189, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.146998594546696e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 27081.25390625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7121221133008896.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6751868724823, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3893728256225586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.422087208126382e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8225467962228736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9417841077911552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.654800999061914e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6280342243508224e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7121221133008896.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.669694185256958, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3891856670379639, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.422087208126382e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.2132428283776222, + "step": 329, + "teaching_rms": 1.2132298119507232, + "traffic_rms": 0.005123541527144432 + }, + { + "batch_loss": 5.204372882843018, + "innovation_rms": 0.6984657323130641, + "instruction_rms": 0.6984601009588121, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2158206007419863e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26879.931640625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6394878275616768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.679168701171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2704479694366455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4284820708327424e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9073073315643392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8476056755372032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.678453555759677e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.365230758966067e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6394878275616768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6736764907836914, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.270228385925293, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4284820708327424e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.69847814463874, + "step": 330, + "teaching_rms": 0.6984657323130641, + "traffic_rms": 0.005159413828791277 + }, + { + "batch_loss": 7.2479753494262695, + "innovation_rms": 0.7694590034720585, + "instruction_rms": 0.769458138017497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.263100975125889e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 26439.263671875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5741105333141504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.682725667953491, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0017136335372925, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4342231536173056e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9835918563213312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7628450865086464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.699740788068188e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1287075542204416e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5741105333141504.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.677233934402466, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0017958879470825, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4342231536173056e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7694749194340567, + "step": 331, + "teaching_rms": 0.7694590034720585, + "traffic_rms": 0.00515263687724699 + }, + { + "batch_loss": 6.125957012176514, + "innovation_rms": 0.76342239269215, + "instruction_rms": 0.7634215750064157, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2856038549778596e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 25749.01171875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5152652937658368.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6858999729156494, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2216097116470337, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4393758258823168e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0522478749155328e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6865605617516544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.718899090987418e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.915836734373888e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5152652937658368.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6804089546203613, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2216919660568237, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4393758258823168e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7634380828713512, + "step": 332, + "teaching_rms": 0.76342239269215, + "traffic_rms": 0.005081960050256244 + }, + { + "batch_loss": 5.977670669555664, + "innovation_rms": 0.927103791469379, + "instruction_rms": 0.9271008725500851, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.28365194696065e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 24802.1328125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4622993812619264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.688730001449585, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9775581359863281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4439987568810394e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.114038280912896e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6179044894703616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.736141494895247e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7242530609364992e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4622993812619264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.683239459991455, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9776403903961182, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4439987568810394e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9271137969824076, + "step": 333, + "teaching_rms": 0.927103791469379, + "traffic_rms": 0.005015180527409155 + }, + { + "batch_loss": 4.730621814727783, + "innovation_rms": 1.1270446560844627, + "instruction_rms": 1.1270395774902942, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2534640305863393e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 23663.36328125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4146254321811456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6912500858306885, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0904147624969482, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.448144946509906e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1696497322098688e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5561140297859072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.751659727131771e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5518277333680128e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4146254321811456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.685760259628296, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0904440879821777, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.448144946509906e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.1270499420409956, + "step": 334, + "teaching_rms": 1.1270446560844627, + "traffic_rms": 0.004955452755976563 + }, + { + "batch_loss": 4.047018051147461, + "innovation_rms": 1.1522882680674469, + "instruction_rms": 1.1522873846906998, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.198003839447597e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 22407.634765625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3717147360493568.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.693491220474243, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0672166347503662, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.451862154805248e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2196999524777984e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5005026321760256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.765626273583596e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3966448741318656e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3717147360493568.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.688002109527588, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0671088695526123, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.451862154805248e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.152297738854542, + "step": 335, + "teaching_rms": 1.1522882680674469, + "traffic_rms": 0.00489558817220692 + }, + { + "batch_loss": 3.5104641914367676, + "innovation_rms": 1.3612572084537276, + "instruction_rms": 1.3612569504993548, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1406744785416356e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 21172.212890625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3330913836466176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.695481300354004, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0027934312820435, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.455192987842642e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2647451399815168e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4504523582210048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.778195753073377e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2569803759812608e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3330913836466176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.689992904663086, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0026856660842896, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.455192987842642e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.3612658409548353, + "step": 336, + "teaching_rms": 1.3612572084537276, + "traffic_rms": 0.004920497919811268 + }, + { + "batch_loss": 4.536196708679199, + "innovation_rms": 1.1568673142634316, + "instruction_rms": 1.1568668271896962, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.101625323081171e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 20040.44140625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2983270593593344.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6972432136535645, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9430446624755859, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4581762721264435e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3052858731593728e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4054071170301952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.789508353333658e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1312823598579712e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2983270593593344.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.691755533218384, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9430726766586304, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4581762721264435e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.1568776528200302, + "step": 337, + "teaching_rms": 1.1568673142634316, + "traffic_rms": 0.005003780651115537 + }, + { + "batch_loss": 3.903050184249878, + "innovation_rms": 1.115619462543129, + "instruction_rms": 1.115616711691902, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0966250190759526e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 19152.759765625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2670361824985088.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.698962450027466, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1659021377563477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4608467109924045e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.341772479332352e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3648663838523392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.799689831006863e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0181541023973376e+16, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2670361824985088.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6934752464294434, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165930151939392, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4608467109924045e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.1156280585308462, + "step": 338, + "teaching_rms": 1.115619462543129, + "traffic_rms": 0.005077874459792762 + }, + { + "batch_loss": 4.258658409118652, + "innovation_rms": 0.929005831079437, + "instruction_rms": 0.9290045912649745, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1248929132702597e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 18539.900390625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2388716928630784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7005956172943115, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4109281301498413, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.463235400003748e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3746104034131968e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3283797239922688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.808852886034842e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9163386814201856.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2388716928630784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6951088905334473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4109561443328857, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.463235400003748e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9290183099565517, + "step": 339, + "teaching_rms": 0.929005831079437, + "traffic_rms": 0.005119375900198495 + }, + { + "batch_loss": 3.5544490814208984, + "innovation_rms": 0.8278712507507313, + "instruction_rms": 0.8278674291883269, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1819812062518313e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 18171.15234375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2135212792217600.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7020437717437744, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3816676139831543, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4653706859446272e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4041646102478848e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2955417462243328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.817099910437929e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8247048025407488.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2135212792217600.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6965577602386475, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3816957473754883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4653706859446272e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8278831971008139, + "step": 340, + "teaching_rms": 0.8278712507507313, + "traffic_rms": 0.005173980067696939 + }, + { + "batch_loss": 3.7154829502105713, + "innovation_rms": 0.8308643602310706, + "instruction_rms": 0.8308612797935344, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2614489589051556e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 18018.96484375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1907037688561664.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.703252077102661, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9007612466812134, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4672776514240512e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4307633427120128e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2659875662331904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.824522301120184e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7422342900744192.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1907037688561664.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6977665424346924, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9005942344665527, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4672776514240512e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8308769840785208, + "step": 341, + "teaching_rms": 0.8308643602310706, + "traffic_rms": 0.005204013959066228 + }, + { + "batch_loss": 4.065582752227783, + "innovation_rms": 0.8688396148864103, + "instruction_rms": 0.8688427071360576, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.348662770976162e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 17997.875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1701661076619264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7043190002441406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1878403425216675, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4689793174667264e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.454702201929728e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2393888069255168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.831202521453691e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6680108664356864.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1701661076619264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.69883394241333, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1876733303070068, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4689793174667264e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8688584973902671, + "step": 342, + "teaching_rms": 0.8688396148864103, + "traffic_rms": 0.005170583604762769 + }, + { + "batch_loss": 4.02751350402832, + "innovation_rms": 0.9288514519783345, + "instruction_rms": 0.9288482584558048, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4601573732012524e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 18299.630859375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1516805147328512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.705252170562744, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.165120840072632, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4704961281169818e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4762471537508352e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2154499208642560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.837214788473324e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6012097905295360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1516805147328512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.699767589569092, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.165116310119629, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4704961281169818e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.928862122357023, + "step": 343, + "teaching_rms": 0.9288514519783345, + "traffic_rms": 0.005142927808215437 + }, + { + "batch_loss": 3.1495614051818848, + "innovation_rms": 0.89925628716287, + "instruction_rms": 0.8992518954560195, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5661016407293297e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 18785.845703125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1350419590676480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7060649394989014, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8506603240966797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.471846465834844e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4956376426020864e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1939049287778304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.842625760071516e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5410887846330368.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1350419590676480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7005810737609863, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8506557941436768, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.471846465834844e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8992662313063784, + "step": 344, + "teaching_rms": 0.89925628716287, + "traffic_rms": 0.0051176283864204745 + }, + { + "batch_loss": 3.198390483856201, + "innovation_rms": 0.9162048225248365, + "instruction_rms": 0.916202637838497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.658398220177637e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 19448.6015625, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1200659181338624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7067620754241943, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3072474002838135, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.473047166892114e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.513089061093376e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1745144265048064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.847495222193029e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4869798739574784.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1200659181338624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7012786865234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3072429895401, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.473047166892114e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9162167541227076, + "step": 345, + "teaching_rms": 0.9162048225248365, + "traffic_rms": 0.005122446695848414 + }, + { + "batch_loss": 3.358837604522705, + "innovation_rms": 0.9807668884644559, + "instruction_rms": 0.9807617610851475, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.737354150168232e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 20270.484375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1065862773604352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.707362413406372, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8856156468391418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4741130059762893e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5287953269981184e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1570629811699712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.851878150419251e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4382818704556032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1065862773604352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7050750255584717, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8854486346244812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4741130059762893e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9807745215734704, + "step": 346, + "teaching_rms": 0.9807668884644559, + "traffic_rms": 0.005086486559686008 + }, + { + "batch_loss": 3.1672120094299316, + "innovation_rms": 0.9325493952199383, + "instruction_rms": 0.9325489137782271, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.793747551801049e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 21098.01953125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 944535316201472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.763143539428711, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8598737716674805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4750575551840256e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.542931030736896e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1413566749999104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.855822648383898e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3944536619352064.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 944535316201472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.767564535140991, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8597067594528198, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4750575551840256e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9325624204673469, + "step": 347, + "teaching_rms": 0.9325493952199383, + "traffic_rms": 0.005019390622966128 + }, + { + "batch_loss": 3.176997184753418, + "innovation_rms": 0.8663215338617005, + "instruction_rms": 0.8663228960680956, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.832408029901095e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 21842.552734375, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 835331142189056.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.802187919616699, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0768303871154785, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4758928404237517e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5556531533643776e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1272209981046784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.85937269655208e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3550082796355584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 835331142189056.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8066084384918213, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0766633749008179, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4758928404237517e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.86633761176817, + "step": 348, + "teaching_rms": 0.8663215338617005, + "traffic_rms": 0.005036069477023855 + }, + { + "batch_loss": 3.7028298377990723, + "innovation_rms": 0.8680103288687832, + "instruction_rms": 0.868011349973286, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.863447792909025e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 22524.423828125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 737039070789632.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9399969577789307, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.378089189529419, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4766298568117453e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5671029993046016e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1144988888989696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.862567465025536e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3195074489876480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 737039070789632.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9444169998168945, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3780848979949951, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4766298568117453e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.868026105642608, + "step": 349, + "teaching_rms": 0.8680103288687832, + "traffic_rms": 0.0050931430414438905 + }, + { + "batch_loss": 3.554030179977417, + "innovation_rms": 0.8077744327429904, + "instruction_rms": 0.8077713302288739, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.878906926395556e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 22990.53125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 648568851398656.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0438220500946045, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.038251280784607, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4772783968734413e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5774079143378944e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1030489959825408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.865443375126938e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2875567040888832.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 648568851398656.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.04824161529541, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0382468700408936, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4772783968734413e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8077875012188334, + "step": 350, + "teaching_rms": 0.8077744327429904, + "traffic_rms": 0.005138505512529699 + }, + { + "batch_loss": 3.2409234046936035, + "innovation_rms": 0.8653150152326538, + "instruction_rms": 0.8653067097844316, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.885059243708776e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 23233.673828125, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568939151097856.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.124375820159912, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8055378198623657, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4778473941408154e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5866823593426944e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 927440943710208.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.868031350620815e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2588010390487040.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568939151097856.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1287949085235596, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8055334091186523, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4778473941408154e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8653216344876623, + "step": 351, + "teaching_rms": 0.8653150152326538, + "traffic_rms": 0.00517447232292819 + }, + { + "batch_loss": 2.6862165927886963, + "innovation_rms": 0.6994373255213455, + "instruction_rms": 0.6994327128574601, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.874316465349591e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 23222.82421875, + "nonfinite_indices": [ + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 497266750128128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1864097118377686, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6203396320343018, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4783445795550003e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5950293061599232e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 834696829206528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.870360253687398e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2329209351438336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 497266750128128.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.190828323364258, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6203352808952332, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4783445795550003e+17, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011661699041724205, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11585475504398346, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.6994511373069514, + "step": 352, + "teaching_rms": 0.6994373255213455, + "traffic_rms": 0.005097079923921986 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} diff --git a/results/kp_innovation_stability_grid/margin_0p01.json b/results/kp_innovation_stability_grid/margin_0p01.json new file mode 100644 index 0000000..506dedb --- /dev/null +++ b/results/kp_innovation_stability_grid/margin_0p01.json @@ -0,0 +1,32986 @@ +{ + "first_any_nonfinite_groups": {}, + "first_any_nonfinite_step": null, + "first_nonfinite_by_group": {}, + "first_nonfinite_by_metric": {}, + "first_nonfinite_metrics": [], + "first_training_failure_groups": {}, + "first_training_failure_step": null, + "initial_parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4011031687259674, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3956821858882904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1162850856781006, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 0.011125171557068825, + "max_applied_stability_margin": 0.0111251724883914, + "max_positive_residual_soma_slope": 0.0, + "min_residual_soma_slope": -0.011125171557068825, + "mse": 8.080039934493075e-05, + "observations": 64, + "residual_traffic_rms_ratio": 0.5331967502228365, + "stability_margin": 0.01 + }, + "first_mse": 8.080039934493075e-05, + "last_mse": 8.080039934493075e-05, + "post_warmup_traffic_residual_rms_ratio": 0.5331967502228365, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_mixed_traffic_nonfinite_diagnosis_v1", + "provenance": { + "git_commit": "0ae690e7b3144da2c5b3384cd7169e86529fd26f", + "git_tracked_dirty": false + }, + "rule": "innovation", + "scope": "training_only_no_validation_or_test_evaluation", + "split": { + "dataset": "cifar10", + "input_layout": "NCHW", + "input_shape": [ + 3, + 32, + 32 + ], + "loader_seed": 0, + "normalization_mean": [ + 0.49140000343322754, + 0.4821999967098236, + 0.4465000033378601 + ], + "normalization_std": [ + 0.24699999392032623, + 0.2434999942779541, + 0.26159998774528503 + ], + "split_from_training_only": true, + "split_seed": 2027, + "test_examples": 10000, + "train_examples": 45000, + "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5", + "validation_class_counts": { + "0": 500, + "1": 500, + "2": 500, + "3": 500, + "4": 500, + "5": 500, + "6": 500, + "7": 500, + "8": 500, + "9": 500 + }, + "validation_examples": 5000, + "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b" + }, + "stability_margin": 0.01, + "test_evaluations": 0, + "traffic_calibration": { + "instruction_rms": [ + 0.008471464738249779, + 0.006617916747927666, + 0.003988116513937712, + 0.0043865954503417015, + 0.002713605063036084, + 0.003158062230795622, + 0.002004207344725728, + 0.003757704282179475, + 0.0027085731271654367, + 0.0030200635083019733, + 0.002056022873148322, + 0.002389939734712243, + 0.0016842411132529378, + 0.0034010226372629404, + 0.002549725119024515, + 0.0030542907770723104, + 0.0020918569061905146, + 0.002605808200314641, + 0.0018916188273578882 + ], + "realized_traffic_instruction_rms_ratio": [ + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 3.999999523162842, + 4.0, + 4.000000476837158, + 3.999999761581421, + 4.0, + 4.0, + 3.999999761581421, + 4.0, + 4.000000476837158, + 4.000000476837158, + 4.0, + 4.0, + 4.0 + ], + "target_ratio": 4.0, + "traffic_gain": [ + 0.0443243607878685, + 0.035817068070173264, + 0.013778294436633587, + 0.022762222215533257, + 0.007214079145342112, + 0.017218582332134247, + 0.004616298712790012, + 0.02042512409389019, + 0.007327491417527199, + 0.01611751690506935, + 0.004703814629465342, + 0.01256294921040535, + 0.003324519144371152, + 0.01805141195654869, + 0.006063441745936871, + 0.016296451911330223, + 0.0044285994954407215, + 0.013874167576432228, + 0.0036007820162922144 + ], + "unscaled_traffic_rms": [ + 0.7644973993301392, + 0.7390796542167664, + 1.157796859741211, + 0.7708553671836853, + 1.504616141319275, + 0.7336404919624329, + 1.736635684967041, + 0.7358984351158142, + 1.478581428527832, + 0.7495108246803284, + 1.7483876943588257, + 0.7609485983848572, + 2.0264477729797363, + 0.7536302804946899, + 1.6820316314697266, + 0.7496824264526367, + 1.8894071578979492, + 0.7512690424919128, + 2.10134220123291 + ] + }, + "trajectory": [ + { + "batch_loss": 2.9134020805358887, + "innovation_rms": 0.010081594228070463, + "instruction_rms": 0.00421539791368719, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40285658836364746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9875131845474243, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.376714706420898, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4058851897716522, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6322960257530212, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9460533857345581, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39024874567985535, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9024875164031982, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0241279602050781, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.568367958068848, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.37673568725586, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3794170618057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.632293164730072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.203450083732605, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01731826256776222, + "step": 1, + "teaching_rms": 0.010081594228070463, + "traffic_rms": 0.016800932580503868 + }, + { + "batch_loss": 2.3531951904296875, + "innovation_rms": 0.0959486766972197, + "instruction_rms": 0.0957219530896388, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6021227836608887, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.693324327468872, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.437236785888672, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40752050280570984, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5040203332901001, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7448183298110962, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7434362173080444, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.591897010803223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.011120319366455, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.987762451171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.437277793884277, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38057780265808105, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5040148496627808, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8815804719924927, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.09608516507643175, + "step": 2, + "teaching_rms": 0.0959486766972197, + "traffic_rms": 0.008262247913726184 + }, + { + "batch_loss": 2.476445436477661, + "innovation_rms": 16.099985356110803, + "instruction_rms": 16.099985018739243, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2091940641403198, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.727959394454956, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.644065856933594, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41007092595100403, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36438536643981934, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.224384784698486, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.47871971130371, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.10513305664062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.049942016601562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.5108871459961, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.59001159667969, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40170571208000183, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36437755823135376, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.532761573791504, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16.099986497611177, + "step": 3, + "teaching_rms": 16.099985356110803, + "traffic_rms": 0.004914635765512741 + }, + { + "batch_loss": 4.717585563659668, + "innovation_rms": 0.03450619975424598, + "instruction_rms": 0.008064917344835476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.71405792236328, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7394.06103515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.278602600097656, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9904522895812988, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.045193672180176, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15224552154541, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.664196014404297, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.42710876464844, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.688008308410645, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.08735656738281, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.22111511230469, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7778841257095337, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.045267105102539, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.15487289428711, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.09613428778531108, + "step": 4, + "teaching_rms": 0.03450619975424598, + "traffic_rms": 0.09580256463310743 + }, + { + "batch_loss": 3.9219820499420166, + "innovation_rms": 0.03796869601488627, + "instruction_rms": 0.030948467456718382, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.6272201538086, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24639.080078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.349918365478516, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.891523003578186, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.010706901550293, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.687237739562988, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.99488067626953, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.7504425048828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.351335525512695, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.95751190185547, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.60025405883789, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.678963541984558, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.010793685913086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.114898681640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.08668192033364577, + "step": 5, + "teaching_rms": 0.03796869601488627, + "traffic_rms": 0.08097444324816802 + }, + { + "batch_loss": 3.992255926132202, + "innovation_rms": 0.13589132503655507, + "instruction_rms": 0.13523142019951254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.2039794921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22438.4375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.05703735351562, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.702415943145752, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.108929634094238, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.86840534210205, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.58065414428711, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 173.16921997070312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.080402374267578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.04670715332031, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.05722045898438, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4898664951324463, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.109028816223145, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.478689193725586, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.13802826740103066, + "step": 6, + "teaching_rms": 0.13589132503655507, + "traffic_rms": 0.027650310279709857 + }, + { + "batch_loss": 3.608405590057373, + "innovation_rms": 0.013078816724299534, + "instruction_rms": 0.006476119296010742, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 105.34825134277344, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 31767.5859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.2530517578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.432016134262085, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2960028648376465, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.731307983398438, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.48666000366211, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.5965576171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.189857482910156, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.73487854003906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.25325393676758, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.219477891921997, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.29611349105835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.30579948425293, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02595734653270197, + "step": 7, + "teaching_rms": 0.013078816724299534, + "traffic_rms": 0.025146052477695365 + }, + { + "batch_loss": 4.016349792480469, + "innovation_rms": 0.02493239233244421, + "instruction_rms": 0.008115198625113972, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.33341979980469, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 49775.22265625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.83855438232422, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.079925060272217, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.479089736938477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.307743072509766, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.20425415039062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.7844696044922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.253345489501953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.9995574951172, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.83851623535156, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.867398977279663, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.47921085357666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.6498908996582, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.030206046833929266, + "step": 8, + "teaching_rms": 0.02493239233244421, + "traffic_rms": 0.02910467753527998 + }, + { + "batch_loss": 4.285613536834717, + "innovation_rms": 0.02372878930757595, + "instruction_rms": 0.006981654984723471, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.53214263916016, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75540.9140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.85408401489258, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.656559944152832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.7663493156433105, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.131980895996094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.75064086914062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.11331176757812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.704914093017578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 166.79315185546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.854042053222656, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.444046974182129, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.766479969024658, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.559104919433594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019655422344281314, + "step": 9, + "teaching_rms": 0.02372878930757595, + "traffic_rms": 0.01838870649647694 + }, + { + "batch_loss": 3.5335745811462402, + "innovation_rms": 0.012893581678316203, + "instruction_rms": 0.0065881682828800876, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.0907211303711, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 103767.53125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.45674133300781, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.17546272277832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.189027309417725, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.596832275390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.24238586425781, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.81564331054688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.96562385559082, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.11383056640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.45669937133789, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.962963581085205, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.189166069030762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.07658767700195, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018757318966569125, + "step": 10, + "teaching_rms": 0.012893581678316203, + "traffic_rms": 0.017571618449407387 + }, + { + "batch_loss": 3.4827187061309814, + "innovation_rms": 0.020468156119700223, + "instruction_rms": 0.011324252957045394, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.9716567993164, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 202009.828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.61726760864258, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.642416477203369, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.669539451599121, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.61492156982422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.20376586914062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.96839141845703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.715017318725586, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.50912857055664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.61722183227539, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.429932117462158, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.6696858406066895, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 45.2414436340332, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.022907930117821326, + "step": 11, + "teaching_rms": 0.020468156119700223, + "traffic_rms": 0.019916931885873533 + }, + { + "batch_loss": 4.213178634643555, + "innovation_rms": 0.03997673600102462, + "instruction_rms": 0.03800572644876113, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.5232162475586, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 265945.28125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.254390716552734, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.060915946960449, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.184996604919434, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.230892181396484, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.7209243774414, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.17154693603516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.990610122680664, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.44788360595703, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.25434112548828, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.848447322845459, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.185149669647217, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.088958740234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03976469849141744, + "step": 12, + "teaching_rms": 0.03997673600102462, + "traffic_rms": 0.011710525315856358 + }, + { + "batch_loss": 3.547884225845337, + "innovation_rms": 0.04959049681590618, + "instruction_rms": 0.0414100338622459, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.97247314453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 240780.734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.326480865478516, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.43750524520874, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7658908367156982, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.4849967956543, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.08636474609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103.65438842773438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.97296905517578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 225.1593017578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.32643508911133, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.225052356719971, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.766049861907959, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.6511344909668, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.045987377786838216, + "step": 13, + "teaching_rms": 0.04959049681590618, + "traffic_rms": 0.02001155970961695 + }, + { + "batch_loss": 4.099434852600098, + "innovation_rms": 0.030805955385111612, + "instruction_rms": 0.02345406785659076, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.50049591064453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1557456.625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.88054275512695, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.773758888244629, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3625361919403076, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.42168045043945, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.4152603149414, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.28894805908203, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.73145294189453, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.84249877929688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.88072204589844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.561322212219238, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3627004623413086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.956565856933594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03904105132729888, + "step": 14, + "teaching_rms": 0.030805955385111612, + "traffic_rms": 0.03121298692662914 + }, + { + "batch_loss": 4.49403715133667, + "innovation_rms": 0.06735415991048833, + "instruction_rms": 0.06529582666987746, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.79850769042969, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1402020.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.99201965332031, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.073694229125977, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9993536472320557, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.069618225097656, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.81491088867188, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.99649810791016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.31565856933594, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.09706115722656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.99220275878906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.861274719238281, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9995226860046387, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.03092575073242, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.07368662364952919, + "step": 15, + "teaching_rms": 0.06735415991048833, + "traffic_rms": 0.03414934946054371 + }, + { + "batch_loss": 4.082787036895752, + "innovation_rms": 0.032925186108314033, + "instruction_rms": 0.024781017940796676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.39004516601562, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1773419.125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.65672302246094, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.3423752784729, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.686810255050659, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.45232391357422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 143.41958618164062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.04668426513672, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.7414436340332, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.2861328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.65672302246094, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1299729347229, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.686983585357666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.89729690551758, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.030638946332581993, + "step": 16, + "teaching_rms": 0.032925186108314033, + "traffic_rms": 0.01802154534716437 + }, + { + "batch_loss": 4.7557692527771, + "innovation_rms": 145.33944056094387, + "instruction_rms": 145.33943888737656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.04461669921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2519415.75, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 434.11712646484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.5841145515441895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4173948764801025, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.59629440307617, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.30726623535156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.87674713134766, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.675392150878906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.01974487304688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 434.1169128417969, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.371730327606201, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.417572021484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.57646179199219, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 145.33944266080783, + "step": 17, + "teaching_rms": 145.33944056094387, + "traffic_rms": 0.033126767453312854 + }, + { + "batch_loss": 4.059061050415039, + "innovation_rms": 4.436981768444155, + "instruction_rms": 4.436955053413805, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.59840393066406, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2320961.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 390.700927734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.801482677459717, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1736795902252197, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.24465942382812, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 156.5061798095703, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.989070892333984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.11990737915039, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.81776428222656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 390.7007141113281, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.58911657333374, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1738603115081787, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48573303222656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.437048537452248, + "step": 18, + "teaching_rms": 4.436981768444155, + "traffic_rms": 0.02880254522379143 + }, + { + "batch_loss": 7.028702735900879, + "innovation_rms": 0.5707960384057782, + "instruction_rms": 0.5706020323831402, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.51519775390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2097844.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.6224365234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.990458011627197, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.373170375823975, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.40690612792969, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.08518981933594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.7901611328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.819969177246094, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.4359893798828, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.6222229003906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.778110027313232, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.373010158538818, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.64795684814453, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.5712966650351449, + "step": 19, + "teaching_rms": 0.5707960384057782, + "traffic_rms": 0.028175225166828882 + }, + { + "batch_loss": 7.875793933868408, + "innovation_rms": 0.2271058321131441, + "instruction_rms": 0.22207768968274055, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.55233764648438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9715103.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 344.1450500488281, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.149005889892578, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.581125020980835, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.0508575439453, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.10630798339844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.211143493652344, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.920204162597656, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 365.2312927246094, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 344.14483642578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.936676979064941, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5809621810913086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.29188537597656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.22514969636042226, + "step": 20, + "teaching_rms": 0.2271058321131441, + "traffic_rms": 0.03707684169613999 + }, + { + "batch_loss": 8.353155136108398, + "innovation_rms": 0.0783114862106525, + "instruction_rms": 0.06808342204504879, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.81932067871094, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778082.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309.7271423339844, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.291596412658691, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.879405975341797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 178.52891540527344, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.62530517578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 264.5400390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.43459701538086, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 328.7081604003906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309.7269287109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.079286575317383, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.879554748535156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 178.76991271972656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.08084429382613317, + "step": 21, + "teaching_rms": 0.0783114862106525, + "traffic_rms": 0.043603594773344745 + }, + { + "batch_loss": 8.006993293762207, + "innovation_rms": 0.06563890372502573, + "instruction_rms": 0.05585668007859487, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.51760864257812, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15499406.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 294.6337585449219, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.4197359085083, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.830373764038086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 204.15737915039062, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 175.69241333007812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.08602905273438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.70216369628906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 131.5056915283203, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 294.6338806152344, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.20744514465332, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.830523490905762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 204.3983612060547, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06191756154683682, + "step": 22, + "teaching_rms": 0.06563890372502573, + "traffic_rms": 0.02674067065547566 + }, + { + "batch_loss": 7.177463531494141, + "innovation_rms": 0.040374080690925795, + "instruction_rms": 0.005132023086861793, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 209.08587646484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 37142528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 265.1800537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.534996032714844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.907163143157959, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 227.220947265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 179.35279846191406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.2774200439453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.15597534179688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257.1410827636719, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 265.1801452636719, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.322725296020508, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.907314777374268, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 227.4619140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04318859118134587, + "step": 23, + "teaching_rms": 0.040374080690925795, + "traffic_rms": 0.04288969857696361 + }, + { + "batch_loss": 8.723922729492188, + "innovation_rms": 2.110580508124699, + "instruction_rms": 2.1105327595176826, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.71315002441406, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 33462276.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.6568145751953, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.63864517211914, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.104203224182129, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.97589111328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.64715576171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.84967041015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.86441040039062, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.38978576660156, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.65689086914062, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.42639446258545, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.10435676574707, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 248.2168426513672, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.1106559901050965, + "step": 24, + "teaching_rms": 2.110580508124699, + "traffic_rms": 0.022839290281816225 + }, + { + "batch_loss": 8.477272033691406, + "innovation_rms": 0.04854714509472903, + "instruction_rms": 0.017244466314942093, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.68807983398438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64694784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.78311157226562, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.731842994689941, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.389216899871826, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.6532897949219, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 185.62393188476562, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 173.564697265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.9020004272461, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.2144775390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.783203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.519612312316895, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.389371871948242, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.89422607421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04314164787736114, + "step": 25, + "teaching_rms": 0.04854714509472903, + "traffic_rms": 0.03955776604945336 + }, + { + "batch_loss": 7.216635227203369, + "innovation_rms": 0.07007175965124711, + "instruction_rms": 0.04919473790585064, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.25401306152344, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 58233340.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 711.973388671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.818131446838379, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.749886989593506, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 283.4603271484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.3030242919922, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156.2311248779297, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.32251739501953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384.0735168457031, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 711.9732666015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.605920791625977, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.750042915344238, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 283.70123291015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06113111947517135, + "step": 26, + "teaching_rms": 0.07007175965124711, + "traffic_rms": 0.03630352563671873 + }, + { + "batch_loss": 7.899499416351318, + "innovation_rms": 0.38761681296120054, + "instruction_rms": 0.3851827597454544, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.9579620361328, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 96634368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858.9400634765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.895745277404785, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.138795375823975, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.5838317871094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.71420288085938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.8861541748047, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.20098114013672, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 490.28436279296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858.93994140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.683554649353027, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.138952255249023, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.82470703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3870710718147605, + "step": 27, + "teaching_rms": 0.38761681296120054, + "traffic_rms": 0.038203675397082124 + }, + { + "batch_loss": 4.937027931213379, + "innovation_rms": 2.7671832026716734, + "instruction_rms": 2.7659740098264276, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.73883056640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142178368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 773.0374145507812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.965508460998535, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.624398231506348, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 312.19287109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.88426208496094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.75308990478516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 105.59813690185547, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 558.0458984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 773.0372924804688, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.753337860107422, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.624556064605713, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 312.4337158203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.7668166986973253, + "step": 28, + "teaching_rms": 2.7671832026716734, + "traffic_rms": 0.06827873057653018 + }, + { + "batch_loss": 4.402019023895264, + "innovation_rms": 0.2195155082531269, + "instruction_rms": 0.21369994660808156, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 228.6546173095703, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 127972888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 695.7169799804688, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.028210639953613, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1613945960998535, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 324.5231018066406, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.84068298339844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.92386627197266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.13654327392578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 639.00146484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 695.7168579101562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.816060066223145, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.161553382873535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 324.763916015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.21610634412383697, + "step": 29, + "teaching_rms": 0.2195155082531269, + "traffic_rms": 0.032169402711184604 + }, + { + "batch_loss": 6.787515640258789, + "innovation_rms": 0.33281327294321084, + "instruction_rms": 0.32876799880718566, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1129.2327880859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 288686272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 626.1217041015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.085094451904297, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.737262725830078, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 335.6170654296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.60145568847656, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130.76242065429688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 143.3428497314453, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.48004150390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 626.12158203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.872963905334473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.737422466278076, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 335.85784912109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.332134013080419, + "step": 30, + "teaching_rms": 0.33281327294321084, + "traffic_rms": 0.047183348495037525 + }, + { + "batch_loss": 5.5442214012146, + "innovation_rms": 0.4912555781977062, + "instruction_rms": 0.49089101483332626, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1106.76171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 272572288.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1226.7703857421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.136213302612305, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.36289381980896, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.7355041503906, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.75120544433594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.81983184814453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 164.22853088378906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 421.63201904296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1226.7705078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.924102783203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3630542755126953, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.87677001953125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.49114790565317035, + "step": 31, + "teaching_rms": 0.4912555781977062, + "traffic_rms": 0.015900015400568776 + }, + { + "batch_loss": 3.944164514541626, + "innovation_rms": 0.7408998999258656, + "instruction_rms": 0.7295190653287797, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1015.7631225585938, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 371731520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1104.0810546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.18210506439209, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.025928258895874, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 402.4450988769531, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.2048797607422, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.40371704101562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.5114288330078, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 897.6857299804688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1104.0811767578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.970015525817871, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0260891914367676, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 402.5863342285156, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7353243364072495, + "step": 32, + "teaching_rms": 0.7408998999258656, + "traffic_rms": 0.09221894603806628 + }, + { + "batch_loss": 27.334157943725586, + "innovation_rms": 6.919107846305785, + "instruction_rms": 6.915842468263619, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1824.998046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 781770112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 993.6495971679688, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.23337459564209, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.058635234832764, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 448.0796813964844, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.38540649414062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.01849365234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.91815185546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1696.5177001953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 993.6497802734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.021306037902832, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.058505535125732, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 448.2209167480469, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.917091993786396, + "step": 33, + "teaching_rms": 6.919107846305785, + "traffic_rms": 0.13153944130677583 + }, + { + "batch_loss": 57.09222412109375, + "innovation_rms": 320.5067592275056, + "instruction_rms": 320.50675463091756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1675.472412109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 703593280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 894.2514038085938, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.27538013458252, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.599193572998047, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 489.1463317871094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 223.7478790283203, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 180.01663208007812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 142.51890563964844, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1526.8658447265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 894.2515869140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.063332557678223, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.5994815826416, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 489.2875671386719, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 320.5067579891325, + "step": 34, + "teaching_rms": 320.5067592275056, + "traffic_rms": 0.04655533606155763 + }, + { + "batch_loss": 28.170204162597656, + "innovation_rms": 411.15256489118724, + "instruction_rms": 411.152551480505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1533.6815185546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 633281280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 804.7840576171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.302918434143066, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.13445281982422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 526.1014404296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.9495849609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 229.02017211914062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 275.2448425292969, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1374.17919921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 804.7842407226562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.09089183807373, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.134742736816406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 526.24267578125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 411.152559150942, + "step": 35, + "teaching_rms": 411.15256489118724, + "traffic_rms": 0.07951965305321788 + }, + { + "batch_loss": 40.54689025878906, + "innovation_rms": 30700.072198084814, + "instruction_rms": 30700.07219778774, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1827.9171142578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 570065344.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 724.25537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.364463806152344, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.719635009765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 574.6162109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 263.4642639160156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.968505859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 398.92095947265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1236.76123046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 724.2555541992188, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.152458190917969, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.719924926757812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 574.52880859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30700.07219791285, + "step": 36, + "teaching_rms": 30700.072198084814, + "traffic_rms": 0.08776080016254979 + }, + { + "batch_loss": 17.797855377197266, + "innovation_rms": 403.5931129836903, + "instruction_rms": 403.5931117737206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2045.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 848048768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 651.7723999023438, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.419178009033203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.214488983154297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 639.79345703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 290.927490234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 377.0716552734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 510.2294616699219, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1113.0850830078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 651.7725219726562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.207193374633789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.214780807495117, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 639.7060546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 403.59311228208264, + "step": 37, + "teaching_rms": 403.5931129836903, + "traffic_rms": 0.020257045830731447 + }, + { + "batch_loss": 27.424907684326172, + "innovation_rms": 729.2354463536564, + "instruction_rms": 729.2354460568572, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2213.6240234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 886117312.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 586.5311889648438, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.618478775024414, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.381807327270508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 698.4465942382812, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 315.6443786621094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339.3644714355469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 610.4071044921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1001.7765502929688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 586.5313110351562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.3204345703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.382101058959961, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 698.3591918945312, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 729.2354464989188, + "step": 38, + "teaching_rms": 729.2354463536564, + "traffic_rms": 0.025419443265906824 + }, + { + "batch_loss": 12.860934257507324, + "innovation_rms": 1031.5539689372276, + "instruction_rms": 1031.553968168041, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2115.095703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 810429568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541.5380859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.987052917480469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.68574333190918, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 751.2274169921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 341.75408935546875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 331.179931640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 700.5670166015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 901.598876953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541.5380249023438, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.689038276672363, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.686037063598633, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 751.1400146484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1031.5539687400435, + "step": 39, + "teaching_rms": 1031.5539689372276, + "traffic_rms": 0.034385276203166545 + }, + { + "batch_loss": 6.751663684844971, + "innovation_rms": 0.2720403663334642, + "instruction_rms": 0.09210476393446367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1904.1337890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2727967488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487.3787841796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.218649864196777, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.315969467163086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 798.72265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 365.9256591796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.0619201660156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 781.7109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2313.211669921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487.37872314453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.92066478729248, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.316264152526855, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 798.63525390625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27344411048185807, + "step": 40, + "teaching_rms": 0.2720403663334642, + "traffic_rms": 0.2574662555454934 + }, + { + "batch_loss": 18.007986068725586, + "innovation_rms": 539688.8008393119, + "instruction_rms": 539688.8008393111, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1721.8265380859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2455171072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438.63055419921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.28950023651123, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.708507537841797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 841.4603881835938, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 387.6800842285156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.5196533203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 854.7404174804688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2081.890380859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438.6304626464844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.991544723510742, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.708802223205566, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 841.3729858398438, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 539688.8008393116, + "step": 41, + "teaching_rms": 539688.8008393119, + "traffic_rms": 0.021584998863311382 + }, + { + "batch_loss": 44.50376892089844, + "innovation_rms": 6237.225529188727, + "instruction_rms": 6237.22553079449, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889.501220703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2260607744.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2938.617431640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.218059539794922, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.285594940185547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 879.9158935546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 407.2590637207031, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 334.0580749511719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 920.4669799804688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.7012939453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2938.61767578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.920133590698242, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.285890579223633, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 879.8285522460938, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6237.225531137563, + "step": 42, + "teaching_rms": 6237.225529188727, + "traffic_rms": 0.0655516204610144 + }, + { + "batch_loss": 64.48465728759766, + "innovation_rms": 12469.311648506515, + "instruction_rms": 12469.311644763031, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4150.80322265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2239953920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2644.726318359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.043323516845703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.652318954467773, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 914.51708984375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 424.880126953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 783.7874145507812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 979.620849609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1794.2752685546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2644.7265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.745427131652832, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.652223587036133, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 914.4297485351562, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12469.311646306822, + "step": 43, + "teaching_rms": 12469.311648506515, + "traffic_rms": 0.19630033762948215 + }, + { + "batch_loss": 36.723934173583984, + "innovation_rms": 82598.21038097251, + "instruction_rms": 82598.21038090314, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4886.62646484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2078369536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2380.19775390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.731266021728516, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.258593559265137, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 945.6489868164062, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 440.7391052246094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 897.5376586914062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1032.859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3403.524169921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2380.1982421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.433399200439453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.258498191833496, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 945.5616455078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 82598.21038096452, + "step": 44, + "teaching_rms": 82598.21038097251, + "traffic_rms": 0.10123789419016044 + }, + { + "batch_loss": 55.940887451171875, + "innovation_rms": 1557.102263389444, + "instruction_rms": 1557.1022594297679, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4397.9638671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870532608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2141.693115234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.29916763305664, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.309737205505371, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010.1687622070312, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 455.0121765136719, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 807.7838745117188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1080.7740478515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3063.171630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2141.693603515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.00132942199707, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.30946159362793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1009.9336547851562, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1557.10226114387, + "step": 45, + "teaching_rms": 1557.102263389444, + "traffic_rms": 0.07314896991308704 + }, + { + "batch_loss": 35.345149993896484, + "innovation_rms": 598.1087116338985, + "instruction_rms": 598.1077044393702, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3948.345458984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1685703936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1926.5615234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.81010627746582, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.726997375488281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1202.824951171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.327392578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 728.788818359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 727.7312622070312, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10629.5302734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1926.5618896484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.512296676635742, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.72672176361084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1202.58984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 598.1079733048625, + "step": 46, + "teaching_rms": 598.1087116338985, + "traffic_rms": 0.5671303034003861 + }, + { + "batch_loss": 133.76547241210938, + "innovation_rms": 158527.87474509006, + "instruction_rms": 158527.87474499148, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3553.510986328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1517133568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1733.994873046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.97441864013672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.88277244567871, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1376.2244873046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 484.4061279296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 655.909912109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 974.443359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9566.5771484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1733.9952392578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.676637649536133, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.882568359375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1375.9893798828125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 158527.87474502518, + "step": 47, + "teaching_rms": 158527.87474509006, + "traffic_rms": 0.10355804713738204 + }, + { + "batch_loss": 6.397861957550049, + "innovation_rms": 10.514262981076765, + "instruction_rms": 10.481370877239348, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3422.409423828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11084286976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1897.5487060546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.122119903564453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.994735717773438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1532.23974609375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 514.4014892578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 590.3189086914062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1196.4842529296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2220.40869140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1897.5484619140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.82436752319336, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.994531631469727, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1532.004638671875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10.492077872904646, + "step": 48, + "teaching_rms": 10.514262981076765, + "traffic_rms": 0.47388392427715326 + }, + { + "batch_loss": 70.169677734375, + "innovation_rms": 414.2098361758039, + "instruction_rms": 414.20911138057807, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3128.46533203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 21260943360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18922.873046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.16143798828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.295331954956055, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1955.88427734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 592.2528076171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 778.5133056640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 799.4066772460938, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4612.81982421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18922.87109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.86371421813965, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.29512882232666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1956.134521484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 414.20930694400096, + "step": 49, + "teaching_rms": 414.2098361758039, + "traffic_rms": 0.4025442846381394 + }, + { + "batch_loss": 131.86611938476562, + "innovation_rms": 1559.9342881903149, + "instruction_rms": 1559.9340093590085, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2815.038330078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24852635648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117730.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.19664192199707, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.264768600463867, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11768.453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 667.5974731445312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 753.4467163085938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 909.3094482421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6833.69580078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117730.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.898948669433594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.264877319335938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11768.2900390625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1559.934104066611, + "step": 50, + "teaching_rms": 1559.9342881903149, + "traffic_rms": 0.5437015759217971 + }, + { + "batch_loss": 223.55084228515625, + "innovation_rms": 30.203763968395158, + "instruction_rms": 30.20322292355191, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2558.408935546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22367657984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105956.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.932937622070312, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.293128967285156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22364.103515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 735.4076538085938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 678.10205078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1034.61572265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6150.326171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105956.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.635272979736328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.293212890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22363.939453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30.203563567185814, + "step": 51, + "teaching_rms": 30.203763968395158, + "traffic_rms": 0.14953460461273335 + }, + { + "batch_loss": 40.07557678222656, + "innovation_rms": 1086119.7085924225, + "instruction_rms": 1086119.7085924055, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5679.556640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20130891776.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95358.609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.695425033569336, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.06001853942871, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31899.96484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 796.4368286132812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 610.2918090820312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1146.8643798828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5535.29345703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95358.609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.397789001464844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.060100555419922, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31899.80078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1086119.7085924202, + "step": 52, + "teaching_rms": 1086119.7085924225, + "traffic_rms": 0.177880450947875 + }, + { + "batch_loss": 59.78083801269531, + "innovation_rms": 415426.7442073996, + "instruction_rms": 415426.7442059848, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9762.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46754992128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85819.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.48148536682129, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.75715160369873, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40481.921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 961.7960205078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1653.592041015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1257.88720703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5431.63818359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85819.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.183879852294922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.757232666015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40481.7578125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 415426.74420637236, + "step": 53, + "teaching_rms": 415426.7442073996, + "traffic_rms": 0.5674636807020302 + }, + { + "batch_loss": 6.703315734863281, + "innovation_rms": 136680.54904510206, + "instruction_rms": 136680.54904436343, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10354.4453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64794472448.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77233.5546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.28876495361328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.181466102600098, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48205.27734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1118.1910400390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1563.950439453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1357.782958984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10650.353515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77233.5546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.99118995666504, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.181547164916992, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48205.11328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 136680.54904453852, + "step": 54, + "teaching_rms": 136680.54904510206, + "traffic_rms": 0.21878493385890838 + }, + { + "batch_loss": 7.703615188598633, + "innovation_rms": 2.9801227720386643, + "instruction_rms": 2.9603710825675247, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9321.42578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 58315042816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69505.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.115144729614258, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.76320743560791, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55155.81640625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1258.946533203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1407.5552978515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1450.0018310546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9585.318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69505.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.817598342895508, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.763289451599121, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55155.65234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.976886915431514, + "step": 55, + "teaching_rms": 2.9801227720386643, + "traffic_rms": 0.3131452517944176 + }, + { + "batch_loss": 2598.345947265625, + "innovation_rms": 2524010.6716852393, + "instruction_rms": 2524010.671685102, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8389.283203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 52483538944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62549.3203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.55067253112793, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273.21875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61410.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1385.62646484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1488.0880126953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2312.680419921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8626.7861328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62549.3203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.516263961791992, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273.2187805175781, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61410.5859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2524010.671685152, + "step": 56, + "teaching_rms": 2524010.6716852393, + "traffic_rms": 0.5048717758003708 + }, + { + "batch_loss": 1448.7550048828125, + "innovation_rms": 881454.267846008, + "instruction_rms": 881454.2678458358, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7550.35498046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 47235186688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56288.24609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.58628463745117, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 240.35610961914062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67039.578125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1499.638427734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2270.071533203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3089.091064453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7764.107421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56288.24609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.5518798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 240.35614013671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67039.4140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 881454.2678459082, + "step": 57, + "teaching_rms": 881454.267846008, + "traffic_rms": 0.3572339315994472 + }, + { + "batch_loss": 1959.366455078125, + "innovation_rms": 3200212.4507345036, + "instruction_rms": 3200212.4507344915, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7174.85595703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 42511667200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50652.71875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.75907135009766, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.72787475585938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 72104.8515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1602.2491455078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2491.345458984375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3787.860595703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6987.6962890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50652.71875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.72467041015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.7279052734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 72104.6875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3200212.4507344975, + "step": 58, + "teaching_rms": 3200212.4507345036, + "traffic_rms": 0.1913220749479154 + }, + { + "batch_loss": 856.2534790039062, + "innovation_rms": 8718.59403720605, + "instruction_rms": 8718.594035615735, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20238.73828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38260498432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45580.234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.49893951416016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.39865112304688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 76662.875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1694.598876953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2371.378173828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4416.75341796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6288.92626953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45580.234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.46453857421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.39869689941406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 76662.7109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8718.594036214286, + "step": 59, + "teaching_rms": 8718.59403720605, + "traffic_rms": 0.10663840972584693 + }, + { + "batch_loss": 1217.882568359375, + "innovation_rms": 9482.582014825799, + "instruction_rms": 9482.582014244454, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 64066.828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 104684822528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41014.54296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.00479888916016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.05860900878906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80764.328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1777.713623046875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2134.240234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4982.7568359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5660.03369140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41014.54296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 97.97040557861328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.0586700439453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80764.1640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9482.582014698144, + "step": 60, + "teaching_rms": 9482.582014825799, + "traffic_rms": 0.09392252932514422 + }, + { + "batch_loss": 757.6437377929688, + "innovation_rms": 484.3608254424905, + "instruction_rms": 484.36054021563257, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 88372.7890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 386953641984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36905.0078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.92920684814453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.24411010742188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 84454.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.516845703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1920.816162109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5492.15966796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5094.0302734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36905.0078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.89482116699219, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.24415588378906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 84454.6640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 484.3606079873077, + "step": 61, + "teaching_rms": 484.3608254424905, + "traffic_rms": 0.2567074636976771 + }, + { + "batch_loss": 260.5519104003906, + "innovation_rms": 2024.5966125198725, + "instruction_rms": 2024.596597513427, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 117278.5390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 381824532480.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33206.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.45436096191406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.25152587890625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 87775.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1989.920654296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1728.7344970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5950.62255859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4584.626953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33206.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.41998291015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.25157928466797, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 87775.2734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2024.596602206375, + "step": 62, + "teaching_rms": 2024.5966125198725, + "traffic_rms": 0.13825244350132954 + }, + { + "batch_loss": 85.90010070800781, + "innovation_rms": 1164.3514446415068, + "instruction_rms": 1164.3514718980607, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 123015.65625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 354087895040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29876.677734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.7257537841797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.71392822265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 90763.1015625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2145.5068359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1613.238525390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6363.23876953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4126.1640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29876.677734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.69137573242188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.71397399902344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 90762.9375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1164.3514772477822, + "step": 63, + "teaching_rms": 1164.3514446415068, + "traffic_rms": 0.11165118469247219 + }, + { + "batch_loss": 59.3968620300293, + "innovation_rms": 10116327.153258527, + "instruction_rms": 10116327.153258527, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 110749.53125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 318679941120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26879.93359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 145.86865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.42896270751953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 93451.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2285.534423828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1518.615966796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6734.59375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3713.547607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26879.93359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 145.8342742919922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.42900085449219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 93450.9296875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10116327.153258527, + "step": 64, + "teaching_rms": 10116327.153258527, + "traffic_rms": 0.10297234087167896 + }, + { + "batch_loss": 716.4456787109375, + "innovation_rms": 95692.92997692744, + "instruction_rms": 95692.92997688858, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 99674.578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 286811947008.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24182.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.94789123535156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.79237365722656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 95869.3515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2411.55908203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1366.75439453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7068.81298828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3342.19287109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24182.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.91351318359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.79241180419922, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 95869.1875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 95692.92997694827, + "step": 65, + "teaching_rms": 95692.92997692744, + "traffic_rms": 0.10737248756990989 + }, + { + "batch_loss": 205.44073486328125, + "innovation_rms": 761030.2552518495, + "instruction_rms": 761030.2552518464, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 89707.1171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 258130755584.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36857.796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.10960388183594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.61711120605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 98044.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2524.9814453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230.0789794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7369.6103515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3007.973388671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36857.796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.07522583007812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.61714935302734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 98044.6640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 761030.2552518513, + "step": 66, + "teaching_rms": 761030.2552518495, + "traffic_rms": 0.08801240725381025 + }, + { + "batch_loss": 247.6087188720703, + "innovation_rms": 149018.1374245758, + "instruction_rms": 149018.1374245557, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 80736.40625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 232317681664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33171.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.44300842285156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.33399963378906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 100001.7734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2627.0615234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1188.77880859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7640.328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2707.176025390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33171.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.40863037109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.33403015136719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 100001.609375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 149018.1374245838, + "step": 67, + "teaching_rms": 149018.1374245758, + "traffic_rms": 0.09170831181844329 + }, + { + "batch_loss": 373.77630615234375, + "innovation_rms": 631615.7781113124, + "instruction_rms": 631615.778111304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 72662.765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 209085906944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29853.78125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.99472045898438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.51718139648438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101762.0234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2718.93359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1245.2667236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7883.97412109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2436.458251953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29853.78125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.96035766601562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.5172119140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101761.859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 631615.7781113103, + "step": 68, + "teaching_rms": 631615.7781113124, + "traffic_rms": 0.08869441008062406 + }, + { + "batch_loss": 43.323360443115234, + "innovation_rms": 10937.70847418805, + "instruction_rms": 10937.708473406581, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 65396.48828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 188177317888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26867.404296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.88949584960938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.94776153564453, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103345.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2801.618408203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1371.9970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8103.25537109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2192.812255859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26867.404296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.85513305664062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.947792053222656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103345.0703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10937.70847372278, + "step": 69, + "teaching_rms": 10937.70847418805, + "traffic_rms": 0.08322259161864114 + }, + { + "batch_loss": 45.68955612182617, + "innovation_rms": 35572772.48210096, + "instruction_rms": 35572772.48210096, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 67072.171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 198735429632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24179.396484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.19296264648438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.03469467163086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104769.0859375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2876.03466796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1234.79736328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8300.6083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1973.531005859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24179.396484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.15859985351562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.03472900390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104768.921875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35572772.48210096, + "step": 70, + "teaching_rms": 35572772.48210096, + "traffic_rms": 0.06997957217143592 + }, + { + "batch_loss": 72.63777160644531, + "innovation_rms": 48961.191689618805, + "instruction_rms": 48961.19168871419, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 60390.96484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 178862030848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21759.947265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.96420288085938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.712406158447266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106049.5078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2943.00927734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1117.19091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8478.2265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1776.1778564453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21759.947265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.92984008789062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.712440490722656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106049.34375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 48961.1916889527, + "step": 71, + "teaching_rms": 48961.191689618805, + "traffic_rms": 0.15454000084123007 + }, + { + "batch_loss": 51.907379150390625, + "innovation_rms": 7365.387748654858, + "instruction_rms": 7365.387747330699, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 54627.41796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 160979025920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19582.224609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.2563934326172, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.921871185302734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107200.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3003.28662109375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1091.9713134765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8638.0830078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1598.56005859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19582.224609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.22203063964844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.92190170288086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107200.6640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7365.387747971434, + "step": 72, + "teaching_rms": 7365.387748654858, + "traffic_rms": 0.09762449570484948 + }, + { + "batch_loss": 5949.06689453125, + "innovation_rms": 33146.21152242152, + "instruction_rms": 33146.211414310215, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 50625.01953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 450093711360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147648.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.48843383789062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.08814239501953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108235.9453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3059.92138671875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13992.5615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8779.6455078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35978.36328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147648.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.45407104492188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.087989807128906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108235.78125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33146.21144910159, + "step": 73, + "teaching_rms": 33146.21152242152, + "traffic_rms": 1.5196473861236441 + }, + { + "batch_loss": 7292.31591796875, + "innovation_rms": 7364546861.849872, + "instruction_rms": 7364546861.849872, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 473977.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3563353800704.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132881.953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.842041015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.57435607910156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109166.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3392.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23140.388671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8907.0517578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26096.916015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132881.953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.80767822265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.574344635009766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109166.3046875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7364546861.849872, + "step": 74, + "teaching_rms": 7364546861.849872, + "traffic_rms": 12.84343339413785 + }, + { + "batch_loss": 259.0882873535156, + "innovation_rms": 1152140.185484792, + "instruction_rms": 1152140.1854803285, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 821333.125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5315254288384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119590.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.1582794189453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.87348937988281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110002.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5351.22802734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30042.3515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9021.716796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23416.017578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119590.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.12393188476562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.873477935791016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110002.6796875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1152140.185538239, + "step": 75, + "teaching_rms": 1152140.185484792, + "traffic_rms": 11.551734564839107 + }, + { + "batch_loss": 840.32421875, + "innovation_rms": 82.05639928977297, + "instruction_rms": 81.88856259601224, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 772696.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14719875284992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220511.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.42019653320312, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.070072174072266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110754.484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8522.5400390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31713.115234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9124.916015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20570.3203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220511.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.38584899902344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.070064544677734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110754.3203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 82.02821089438132, + "step": 76, + "teaching_rms": 82.05639928977297, + "traffic_rms": 4.784475384170825 + }, + { + "batch_loss": 1997.2177734375, + "innovation_rms": 208858440.19332775, + "instruction_rms": 208858440.19332775, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 695426.6875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13247887441920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198458.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.41690063476562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.338470458984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111429.8515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11376.720703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28541.802734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9217.794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18513.287109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198458.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.38255310058594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.33846664428711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111429.6875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 208858440.19332775, + "step": 77, + "teaching_rms": 208858440.19332775, + "traffic_rms": 0.14988704158466135 + }, + { + "batch_loss": 4261.8828125, + "innovation_rms": 28913118.469627153, + "instruction_rms": 28913118.469627146, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 625884.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11923098697728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178608.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.00485229492188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.30683898925781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112036.5703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13945.482421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25687.62109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9301.3857421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16661.95703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178608.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.9705047607422, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.30693054199219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112036.40625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28913118.46962715, + "step": 78, + "teaching_rms": 28913118.469627153, + "traffic_rms": 0.19705822829599642 + }, + { + "batch_loss": 1241.7724609375, + "innovation_rms": 7600604.911146745, + "instruction_rms": 7600604.911146743, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 563295.625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10730789142528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160741.328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.6320037841797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.23587799072266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112581.4921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16257.3681640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23118.857421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9376.6171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14995.7607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160741.328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.59765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.23606872558594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112581.328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7600604.911146744, + "step": 79, + "teaching_rms": 7600604.911146745, + "traffic_rms": 0.08673790305502886 + }, + { + "batch_loss": 1902.4310302734375, + "innovation_rms": 2883219.6745009255, + "instruction_rms": 2883219.6745009176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 506966.0625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9657710018560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144659.609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.13165283203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.792877197265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113070.796875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18338.064453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20806.970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9444.326171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13496.1845703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144659.609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.09732055664062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.79306411743164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113070.6328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2883219.6745009194, + "step": 80, + "teaching_rms": 2883219.6745009255, + "traffic_rms": 0.09416903355957683 + }, + { + "batch_loss": 844.3853759765625, + "innovation_rms": 266091.74697477254, + "instruction_rms": 266091.7469746488, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 456269.46875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691939016704.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130184.6171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.6126708984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.333309173583984, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113510.0390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20210.69140625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18726.2734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9505.263671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12146.5654296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130184.6171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.57833862304688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.33349609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113509.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 266091.7469746452, + "step": 81, + "teaching_rms": 266091.74697477254, + "traffic_rms": 0.09631282162521156 + }, + { + "batch_loss": 1558.805419921875, + "innovation_rms": 1599338.45800607, + "instruction_rms": 1599338.4580060672, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 410642.53125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7822745010176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117155.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.1436004638672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.66987609863281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115032.328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21896.056640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16853.646484375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9560.107421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10931.908203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117155.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.10926818847656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.66976547241211, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115032.359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1599338.4580060686, + "step": 82, + "teaching_rms": 1599338.45800607, + "traffic_rms": 0.06482730943630734 + }, + { + "batch_loss": 3305.1689453125, + "innovation_rms": 75001.28440809395, + "instruction_rms": 75001.28440800995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 369578.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7040470614016.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105428.7421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.48373413085938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.49239349365234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 125575.203125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 23412.884765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15168.28125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9609.466796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9838.716796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105428.7421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.4494171142578, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.49227905273438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 125575.234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 75001.2844080289, + "step": 83, + "teaching_rms": 75001.28440809395, + "traffic_rms": 0.054547511758923764 + }, + { + "batch_loss": 672.3721923828125, + "innovation_rms": 265932.49983023526, + "instruction_rms": 265932.49983020505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 332620.4375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6336423657472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94873.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.88787841796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.38472747802734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 135062.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 24778.029296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13651.453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10275.8466796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8854.8447265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94873.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.85357666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.38461303710938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 135062.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 265932.4998302202, + "step": 84, + "teaching_rms": 265932.49983023526, + "traffic_rms": 0.0908454686000366 + }, + { + "batch_loss": 2271.1162109375, + "innovation_rms": 4325618.501894768, + "instruction_rms": 4325618.5018947655, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 299358.40625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5702781239296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85372.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.32516479492188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48123168945312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 143599.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 26006.66015625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12286.3076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11072.7822265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7969.35986328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85372.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.29086303710938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48112487792969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 143599.8125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4325618.501894766, + "step": 85, + "teaching_rms": 4325618.501894768, + "traffic_rms": 0.06675792938215408 + }, + { + "batch_loss": 152.3937530517578, + "innovation_rms": 1856.59201868768, + "instruction_rms": 1856.592017309222, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 269422.5625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5132503220224.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76820.859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.81675720214844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.79853820800781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 151281.875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 27112.427734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11057.6767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11790.0244140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7172.423828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76820.859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.78245544433594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.79843139648438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 151281.90625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1856.5920181130946, + "step": 86, + "teaching_rms": 1856.59201868768, + "traffic_rms": 0.05401767380816191 + }, + { + "batch_loss": 327.6507568359375, + "innovation_rms": 24135.696659637317, + "instruction_rms": 24135.696659599616, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 242480.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4619253055488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69123.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.35723876953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.16736602783203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158194.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 28107.619140625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9951.9091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12435.54296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6455.18115234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69123.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.32293701171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.1672592163086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158194.265625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24135.69665967624, + "step": 87, + "teaching_rms": 24135.696659637317, + "traffic_rms": 0.04325624685433459 + }, + { + "batch_loss": 203.13204956054688, + "innovation_rms": 2869.9443332288215, + "instruction_rms": 2869.944332846345, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 218232.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4157327802368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62195.4609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.94171142578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.48736572265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 164413.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29003.291015625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8956.7177734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13016.509765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5809.6630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62195.4609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.90740966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.48725891113281, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 164413.8125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2869.9443330659496, + "step": 88, + "teaching_rms": 2869.9443332288215, + "traffic_rms": 0.03428959260955684 + }, + { + "batch_loss": 39.23117446899414, + "innovation_rms": 735.585376497103, + "instruction_rms": 735.5853756846246, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 196409.046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3741594943488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55959.47265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.5657958984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.63496780395508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 170009.734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29809.396484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8061.0458984375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13539.3798828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5228.69677734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55959.47265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.531494140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.63486099243164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 170009.765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 735.5853761796513, + "step": 89, + "teaching_rms": 735.585376497103, + "traffic_rms": 0.026819681747595088 + }, + { + "batch_loss": 14.122994422912598, + "innovation_rms": 336.7656642176396, + "instruction_rms": 336.7656633965037, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 176768.140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3367435501568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50346.5234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.22552490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.367286682128906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 175044.390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30534.890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7254.94091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14009.962890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4705.8271484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50346.5234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.19122314453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.3671760559082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 175044.421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 336.7656640594655, + "step": 90, + "teaching_rms": 336.7656642176396, + "traffic_rms": 0.020985345841795084 + }, + { + "batch_loss": 17.963912963867188, + "innovation_rms": 3172.4900378929187, + "instruction_rms": 3172.490037850651, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 159091.328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3030691872768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45294.3671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.91734313964844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.62590026855469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 179573.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31187.8359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6529.44677734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14433.4873046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4235.244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45294.3671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.88304138183594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.625789642333984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 179573.859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3172.490037892819, + "step": 91, + "teaching_rms": 3172.4900378929187, + "traffic_rms": 0.016208204895512557 + }, + { + "batch_loss": 34798.5078125, + "innovation_rms": 56.21518078935135, + "instruction_rms": 55.73845907678787, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3799450.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2636418966880256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103991.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.63803100585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.35667419433594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 183648.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31775.486328125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5876.501953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14787.1767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24900.791015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103991.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.60372924804688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.35675048828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 183648.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 55.84365332084459, + "step": 92, + "teaching_rms": 56.21518078935135, + "traffic_rms": 3.4276971823622397 + }, + { + "batch_loss": 1450.1468505859375, + "innovation_rms": 55.555337740404646, + "instruction_rms": 55.53033899674401, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3813598.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5335226091831296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93591.4375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.43551635742188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 224.07655334472656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 187314.390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32308.556640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5330.7109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15097.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22125.76953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93591.4375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.40122985839844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 224.0766143798828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 187314.421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 55.543188036364036, + "step": 93, + "teaching_rms": 55.555337740404646, + "traffic_rms": 1.1943028390434307 + }, + { + "batch_loss": 5330.20166015625, + "innovation_rms": 2014177.681224092, + "instruction_rms": 2014177.6812233496, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432238.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4801703643709440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1584325.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.25132751464844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.12664794921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 190611.796875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32793.953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4853.951171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15303.646484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19913.19140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1584325.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.217041015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.126708984375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 190611.828125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2014177.681223577, + "step": 94, + "teaching_rms": 2014177.681224092, + "traffic_rms": 0.9577668411703144 + }, + { + "batch_loss": 13750.9228515625, + "innovation_rms": 369644.0973330957, + "instruction_rms": 369644.09733159625, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3089014.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4321533279338496.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2592381.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.62945556640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.9471893310547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 301019.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33230.80859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4368.55615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15488.9541015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17921.87109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2592381.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.59518432617188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.94703674316406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 301019.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 369644.0973322402, + "step": 95, + "teaching_rms": 369644.0973330957, + "traffic_rms": 0.6940779991497632 + }, + { + "batch_loss": 4439.3671875, + "innovation_rms": 1629.6140078430903, + "instruction_rms": 1629.609929165954, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11858387.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8349087321554944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2333117.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.97500610351562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.18533325195312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 492549.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33623.98046875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10156.64453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15604.1533203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29920.73046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2333117.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.9407501220703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.18511962890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 492549.6875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1629.620637637452, + "step": 96, + "teaching_rms": 1629.6140078430903, + "traffic_rms": 5.908032157801106 + }, + { + "batch_loss": 39616.0625, + "innovation_rms": 4486436.411854939, + "instruction_rms": 4486436.41185492, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10672548.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7514178428338176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2099756.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.392333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 641.773681640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 702525.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33977.83203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20556.421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15674.3740234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26928.65625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2099756.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.3580780029297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 641.7738037109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 702525.3125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4486436.411854932, + "step": 97, + "teaching_rms": 4486436.411854939, + "traffic_rms": 0.3419906995368841 + }, + { + "batch_loss": 29161.96875, + "innovation_rms": 154396942.94943586, + "instruction_rms": 154396942.94943586, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9605293.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6762760478130176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889710.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.8644256591797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 540.0814208984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 891496.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34296.30078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18500.779296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15737.572265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24235.7890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889710.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.83016967773438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 540.08154296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 891496.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 154396942.94943586, + "step": 98, + "teaching_rms": 154396942.94943586, + "traffic_rms": 0.19069054955236614 + }, + { + "batch_loss": 72121.9921875, + "innovation_rms": 3874518777.135761, + "instruction_rms": 3874518777.135761, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8644764.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6086484591378432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1700650.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.27090454101562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.2171630859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1061561.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34582.921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16650.701171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15794.451171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21812.208984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1700650.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.23666381835938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.2172241210938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1061561.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3874518777.135761, + "step": 99, + "teaching_rms": 3874518777.135761, + "traffic_rms": 0.12258361498714034 + }, + { + "batch_loss": 26738.458984375, + "innovation_rms": 1705363243.6050854, + "instruction_rms": 1705363243.6050854, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7780287.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5477836185927680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530479.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.6424102783203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 705.591796875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1214609.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34840.87890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89920.078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15845.642578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19630.98828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530479.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.48326110839844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 705.5918579101562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1214609.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1705363243.6050854, + "step": 100, + "teaching_rms": 1705363243.6050854, + "traffic_rms": 0.23018973295536913 + }, + { + "batch_loss": 21163.857421875, + "innovation_rms": 45894576.22109617, + "instruction_rms": 45894576.22109617, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7002258.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4930052298899456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1377309.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.15792846679688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 612.1289672851562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1352340.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 90535.5234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 905355.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15891.71484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17667.888671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1377309.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.998779296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 612.1290283203125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1352340.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 45894576.22109617, + "step": 101, + "teaching_rms": 45894576.22109617, + "traffic_rms": 0.21639849755658833 + }, + { + "batch_loss": 2820.987060546875, + "innovation_rms": 17.1334560491678, + "instruction_rms": 2.1544434067781384, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 89562480.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1193850255701443e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1239443.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.03269958496094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 548.3017578125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1476284.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 166117.515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 755819.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15933.1796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15901.099609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1239443.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.92633056640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 548.3018188476562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1476284.625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31.296822141660023, + "step": 102, + "teaching_rms": 17.1334560491678, + "traffic_rms": 31.22258063099652 + }, + { + "batch_loss": 190.3369598388672, + "innovation_rms": 31.177775607304177, + "instruction_rms": 0.057912260984576636, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 392813888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4611736576468713e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1115351.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.3778533935547, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 493.4515686035156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1587819.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 223315.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 571983.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15970.498046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14310.9892578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1115351.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.27149963378906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 493.4516296386719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1587819.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 57.2870409060788, + "step": 103, + "teaching_rms": 31.177775607304177, + "traffic_rms": 57.28701508518365 + }, + { + "batch_loss": 21623.69921875, + "innovation_rms": 10349250224.125568, + "instruction_rms": 10349250224.125568, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 952185408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.316665714270798e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1003657.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 293.01690673828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 436.39056396484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1688185.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 260241.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 369255.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16004.083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12879.8896484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1003657.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 292.9105529785156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 436.3905944824219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1688185.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10349250224.125568, + "step": 104, + "teaching_rms": 10349250224.125568, + "traffic_rms": 77.02182068991524 + }, + { + "batch_loss": 654.7659301757812, + "innovation_rms": 1160356.4762901121, + "instruction_rms": 1160356.4752604428, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1742342528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6622969183359363e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 903122.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 332.28912353515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 392.7221984863281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1778497.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 276514.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162736.390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16034.3115234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18680.341796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 903122.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 332.1827697753906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 392.72222900390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1778497.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1160356.4787314876, + "step": 105, + "teaching_rms": 1160356.4762901121, + "traffic_rms": 89.75131591670393 + }, + { + "batch_loss": 28.654651641845703, + "innovation_rms": 614759.5286315297, + "instruction_rms": 614759.5264401266, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2688386560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.246996815092279e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 812632.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 367.63079833984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 353.416748046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859761.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 273141.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47787.203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16281.65234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24178.984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 812632.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 367.5244445800781, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 353.4167785644531, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859761.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 614759.5338360644, + "step": 106, + "teaching_rms": 614759.5286315297, + "traffic_rms": 95.3595642139492 + }, + { + "batch_loss": 67.26221466064453, + "innovation_rms": 992347.4808665419, + "instruction_rms": 992347.4795422661, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3685697792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.158950075084057e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 731183.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 399.43463134765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 318.0382995605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1932879.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 252305.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 208361.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17220.595703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26787.521484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 731183.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 399.3282775878906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 318.0383605957031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1932879.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 992347.4840128833, + "step": 107, + "teaching_rms": 992347.4808665419, + "traffic_rms": 94.19560326182284 + }, + { + "batch_loss": 22.603679656982422, + "innovation_rms": 817722.773814597, + "instruction_rms": 817722.772443721, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4619218432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.00868962466092e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 657871.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 428.0540771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286.19451904296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1998666.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 217110.515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351947.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18065.64453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24108.76953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 657871.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 427.9477233886719, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286.194580078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1998666.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 817722.7770727874, + "step": 108, + "teaching_rms": 817722.773814597, + "traffic_rms": 87.00911365584236 + }, + { + "batch_loss": 35.365901947021484, + "innovation_rms": 188064.4477042057, + "instruction_rms": 188064.44329043356, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5380583424.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.395158658006458e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 591884.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 453.80731201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257.5322570800781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2057855.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 171286.703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 458238.125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18826.189453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21697.892578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 591884.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 453.7009582519531, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257.5323181152344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2057855.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 188064.45819455187, + "step": 109, + "teaching_rms": 188064.4477042057, + "traffic_rms": 74.87235363678894 + }, + { + "batch_loss": 76704.328125, + "innovation_rms": 791882062.3084093, + "instruction_rms": 791882062.3084087, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5880616448.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.0584945508093e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8878180.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 474.169677734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.87893676757812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2111104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 118882.9296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 524037.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19510.6796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19528.103515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8878180.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 474.0633239746094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.87892150878906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2111104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 791882062.3084109, + "step": 110, + "teaching_rms": 791882062.3084093, + "traffic_rms": 59.068752864613934 + }, + { + "batch_loss": 133819.8125, + "innovation_rms": 3404781711.383887, + "instruction_rms": 3404781711.383887, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6057617408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.970769355880215e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7990273.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 486.6067199707031, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1960.5316162109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2159007.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 64523.9609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 549107.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20126.720703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17575.29296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7990273.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 486.5003662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1960.53173828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2159007.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3404781711.383887, + "step": 111, + "teaching_rms": 3404781711.383887, + "traffic_rms": 41.0166035170652 + }, + { + "batch_loss": 139979.8125, + "innovation_rms": 21167066894.381527, + "instruction_rms": 21167066894.381527, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5885148672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.336531642482279e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191077.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 494.7956237792969, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1908.21337890625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2405952.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 67063.5703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 535885.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20681.158203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15817.763671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191077.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 494.68927001953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1908.21337890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2405953.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 21167066894.381527, + "step": 112, + "teaching_rms": 21167066894.381527, + "traffic_rms": 22.143065772060186 + }, + { + "batch_loss": 1747.673095703125, + "innovation_rms": 677463190.7983509, + "instruction_rms": 677463190.7983509, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5369628672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.510435659146267e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6471728.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 502.1606750488281, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1717.375244140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3053125.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 69349.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 489063.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21180.15234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14235.9873046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6471728.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 502.0543212890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1717.375244140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3053125.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 677463190.7983509, + "step": 113, + "teaching_rms": 677463190.7983509, + "traffic_rms": 4.032699681702085 + }, + { + "batch_loss": 1819.5068359375, + "innovation_rms": 3681439843060.2715, + "instruction_rms": 3681439843060.2715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4832665600.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.759392163600385e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5824250.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 572.256103515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1545.603759765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3635550.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 82538.359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 440157.03125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21629.24609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12812.388671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5824250.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 572.0428466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1545.603759765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3635550.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3681439843060.2715, + "step": 114, + "teaching_rms": 3681439843060.2715, + "traffic_rms": 1.4512551612798417 + }, + { + "batch_loss": 9276.65234375, + "innovation_rms": 456866884546827.94, + "instruction_rms": 456866884546827.94, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4349399040.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.083453228715323e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5241461.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 676.1685180664062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1390.9940185546875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4159696.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 122152.4921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 396141.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22033.4296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11531.1494140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5241461.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 675.9552612304688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1390.9940185546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4159697.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 456866884546827.94, + "step": 115, + "teaching_rms": 456866884546827.94, + "traffic_rms": 0.8427960906421251 + }, + { + "batch_loss": 7904.04150390625, + "innovation_rms": 1492190030155.8389, + "instruction_rms": 1492190030155.8389, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3914457344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.475108046581279e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4716899.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.6829223632812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1251.831298828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4631386.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 157805.203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 356527.15625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22397.1953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10378.0341796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4716899.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.4696655273438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1251.831298828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4631387.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1492190030155.8389, + "step": 116, + "teaching_rms": 1492190030155.8389, + "traffic_rms": 0.34813901344231035 + }, + { + "batch_loss": 40475.6640625, + "innovation_rms": 106669291.61713824, + "instruction_rms": 106669291.61713824, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3523007232.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.9275973474762673e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4244746.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 870.6068725585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1126.5723876953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5055861.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 189892.640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 320874.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22724.583984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9340.23046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4244746.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 870.6082763671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1126.5723876953125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5055861.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 106669291.61713824, + "step": 117, + "teaching_rms": 106669291.61713824, + "traffic_rms": 0.3382242943798017 + }, + { + "batch_loss": 11094.6171875, + "innovation_rms": 64551.29359661899, + "instruction_rms": 64551.29359557233, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3170694400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.434837753466129e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3819765.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 971.9896850585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1013.8280639648438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5437837.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 218771.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288787.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23019.234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8406.20703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3819765.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 971.9910888671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1013.8280639648438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5437838.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 64551.29359614055, + "step": 118, + "teaching_rms": 64551.29359661899, + "traffic_rms": 0.2715866267992134 + }, + { + "batch_loss": 3310.7578125, + "innovation_rms": 23209.36942110678, + "instruction_rms": 23209.369419991126, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2853601536.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9913538725664e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3437245.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063.2244873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 912.3480224609375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5781562.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 244762.171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 259908.296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23284.419921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7565.5859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3437245.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063.2259521484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 912.3480224609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5781562.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23209.36942155309, + "step": 119, + "teaching_rms": 23209.36942110678, + "traffic_rms": 0.2703935283034405 + }, + { + "batch_loss": 2535.720703125, + "innovation_rms": 67877.8695883476, + "instruction_rms": 67877.86958775036, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2568225792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.59221848530976e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3092942.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1145.3251953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 821.0068969726562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6090856.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 268153.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 233917.46875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23523.0859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6809.02734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3092942.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1145.32666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 821.0068969726562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6090857.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 67877.86958831549, + "step": 120, + "teaching_rms": 67877.8695883476, + "traffic_rms": 0.27581190291462626 + }, + { + "batch_loss": 445.22064208984375, + "innovation_rms": 13070.20080237828, + "instruction_rms": 13070.200801421683, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2311390208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.232996601594412e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2783039.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1219.204345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 738.7916259765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6369160.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 289206.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210525.71875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23737.88671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6128.12451171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2783039.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1219.205810546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 738.7916259765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6369161.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13070.200803096188, + "step": 121, + "teaching_rms": 13070.20080237828, + "traffic_rms": 0.2092501311182215 + }, + { + "batch_loss": 4145.38037109375, + "innovation_rms": 3321861.5081997146, + "instruction_rms": 3321861.5081997016, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2080240128.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9096969062505985e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2504098.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1285.683349609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 664.79052734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6619570.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 308153.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 189473.140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23931.20703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5515.31201171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2504098.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1285.684814453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 664.79052734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6619571.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3321861.5081997057, + "step": 122, + "teaching_rms": 3321861.5081997146, + "traffic_rms": 0.18719239052925485 + }, + { + "batch_loss": 30927.05859375, + "innovation_rms": 27850535.482121, + "instruction_rms": 27850535.48212099, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872207744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6187272508099107e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2253026.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1345.5015869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 598.1829223632812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6844873.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 325206.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170525.828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24105.1953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4963.78076171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2253026.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1345.5030517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 598.1829223632812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6844873.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 27850535.48212099, + "step": 123, + "teaching_rms": 27850535.482121, + "traffic_rms": 0.21867303397486074 + }, + { + "batch_loss": 23206.380859375, + "innovation_rms": 2956950822.285266, + "instruction_rms": 2956950822.285266, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1684978688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3568545433211057e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2027039.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1399.3245849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 538.2301025390625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7047577.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 340553.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153473.234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24261.78515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4503.40625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2027039.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1399.3260498046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 538.2301025390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7047577.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2956950822.285266, + "step": 124, + "teaching_rms": 2956950822.285266, + "traffic_rms": 0.20347581446332308 + }, + { + "batch_loss": 49260.859375, + "innovation_rms": 3116872821.463919, + "instruction_rms": 3116872821.463919, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1516471936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.121169036212437e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1823630.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1447.7513427734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 532.6173095703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7229940.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 354366.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 138125.90625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24402.71484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4181.15966796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1823630.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1447.7528076171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 532.6171264648438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7229940.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3116872821.463919, + "step": 125, + "teaching_rms": 3116872821.463919, + "traffic_rms": 0.16762426641332961 + }, + { + "batch_loss": 37775.8125, + "innovation_rms": 3726014308.5505114, + "instruction_rms": 3726014308.5505114, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1364815872.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9090521325911933e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1640544.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1491.3209228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 444.3330078125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7393994.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 366797.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124313.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24529.552734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3618.595703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1640544.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1491.3223876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 444.33282470703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7393995.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3726014308.5505114, + "step": 126, + "teaching_rms": 3726014308.5505114, + "traffic_rms": 0.1545241170647442 + }, + { + "batch_loss": 22684.005859375, + "innovation_rms": 912923143.828508, + "instruction_rms": 912923143.828508, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1228326016.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7181468313711437e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1475750.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530.5186767578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 391.9769287109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7541569.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 377985.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111881.9765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24643.70703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3256.736083984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1475750.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530.5201416015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 391.9769287109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7541570.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 912923143.828508, + "step": 127, + "teaching_rms": 912923143.828508, + "traffic_rms": 0.190676409287617 + }, + { + "batch_loss": 11248.2392578125, + "innovation_rms": 844098271.448619, + "instruction_rms": 844098271.448619, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1105486208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5463322186027735e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1327421.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1565.78125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 352.6261901855469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7674311.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 388055.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100693.7734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24746.4453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2931.0625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1327421.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1565.78271484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 352.6261901855469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7674312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 844098271.448619, + "step": 128, + "teaching_rms": 844098271.448619, + "traffic_rms": 0.20476248185497917 + }, + { + "batch_loss": 6286.47802734375, + "innovation_rms": 1530695490.1888525, + "instruction_rms": 1530695490.1888525, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 994931840.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3916990143346822e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193911.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1597.501953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 317.20697021484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7793702.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 397117.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90624.390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24838.91015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2637.956298828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193911.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1597.50341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 317.20697021484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7793703.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1530695490.1888525, + "step": 129, + "teaching_rms": 1530695490.1888525, + "traffic_rms": 0.21808496229076418 + }, + { + "batch_loss": 4260.3583984375, + "innovation_rms": 1937043802.3431761, + "instruction_rms": 1937043802.3431761, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 895434560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.252529077716842e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1073741.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1626.0345458984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.3265075683594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7901076.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 405273.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81561.953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24922.12890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2374.16064453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1073741.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1626.0360107421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.3265075683594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7901077.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1937043802.3431761, + "step": 130, + "teaching_rms": 1937043802.3431761, + "traffic_rms": 0.21713706529412918 + }, + { + "batch_loss": 6951.6337890625, + "innovation_rms": 3142017317.2697797, + "instruction_rms": 3142017317.2697797, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 805888768.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1272761435568787e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 965576.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1651.6976318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256.6312561035156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7997634.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 412614.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73405.7578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24997.025390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2136.74462890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 965576.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1651.6990966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256.6312561035156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7997634.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3142017317.2697797, + "step": 131, + "teaching_rms": 3142017317.2697797, + "traffic_rms": 0.19990978667274367 + }, + { + "batch_loss": 9759.3134765625, + "innovation_rms": 18161820896.946182, + "instruction_rms": 18161820896.946182, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 725299328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0145484940168187e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868219.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1674.7779541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.80294799804688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8084456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 419220.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66065.1796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25064.431640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1923.070068359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868219.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1674.7794189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.80294799804688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8084456.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 18161820896.946182, + "step": 132, + "teaching_rms": 18161820896.946182, + "traffic_rms": 0.17481659425787535 + }, + { + "batch_loss": 17075.978515625, + "innovation_rms": 245947640.18787143, + "instruction_rms": 245947640.18787143, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 652769408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.130936534112299e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 780588.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1695.533447265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.55516052246094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8162515.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 425166.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59458.66015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25125.09765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730.7630615234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 780588.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1695.534912109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.55516052246094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8162515.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 245947640.18787143, + "step": 133, + "teaching_rms": 245947640.18787143, + "traffic_rms": 0.20005679938945128 + }, + { + "batch_loss": 157.33346557617188, + "innovation_rms": 1878873.1755825013, + "instruction_rms": 1878873.1755824971, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 587492480.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.21784314458386e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701713.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1714.1964111328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.6300811767578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8232686.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 430518.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53512.79296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25179.697265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1557.686767578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701713.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1714.1978759765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.6300811767578, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8232687.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1878873.1755825072, + "step": 134, + "teaching_rms": 1878873.1755825013, + "traffic_rms": 0.19276152983768027 + }, + { + "batch_loss": 626.1778564453125, + "innovation_rms": 28219708.03182532, + "instruction_rms": 28219708.03182532, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 528743232.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.396058478281753e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 630719.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730.9759521484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.795654296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8295758.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 435334.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48161.51171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25228.8359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1401.9180908203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 630719.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730.9774169921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.795654296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8295759.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28219708.03182532, + "step": 135, + "teaching_rms": 28219708.03182532, + "traffic_rms": 0.17212384620494095 + }, + { + "batch_loss": 79797.4609375, + "innovation_rms": 23822876.746210117, + "instruction_rms": 23822876.746210117, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 475868896.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.656452630453577e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 566817.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1746.060302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311.909423828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8352440.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 439668.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43345.359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25273.060546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1261.7261962890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 566817.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1746.061767578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311.90960693359375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8352441.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23822876.746210117, + "step": 136, + "teaching_rms": 23822876.746210117, + "traffic_rms": 0.1044686088665963 + }, + { + "batch_loss": 107035.4140625, + "innovation_rms": 56.24663941904081, + "instruction_rms": 56.24183297400059, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 428282016.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.990807411388685e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 509300.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1759.61865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.5791015625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8403371.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 443569.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39010.82421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25312.86328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1135.5535888671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 509300.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1759.6201171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.57928466796875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8403371.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 56.25104374635909, + "step": 137, + "teaching_rms": 56.24663941904081, + "traffic_rms": 1.0181515222660913 + }, + { + "batch_loss": 87151.8203125, + "innovation_rms": 41673.73587703515, + "instruction_rms": 41673.735856783984, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 385453824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.391726626269351e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 457530.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1771.8035888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.0955047607422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8449124.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 447080.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35109.7421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25348.685546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1235.1380615234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 457530.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1771.8050537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.09568786621094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8449124.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41673.73589848967, + "step": 138, + "teaching_rms": 41673.73587703515, + "traffic_rms": 1.8645443710527563 + }, + { + "batch_loss": 74592.7109375, + "innovation_rms": 22658.386580346014, + "instruction_rms": 22658.385531534484, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 346908448.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.852553831701021e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410932.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1782.7523193359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 326.108154296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490217.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 450240.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31598.767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25380.92578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1471.967529296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410932.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1782.7537841796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 326.108154296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490217.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22658.38567014104, + "step": 139, + "teaching_rms": 22658.386580346014, + "traffic_rms": 2.5062198463671486 + }, + { + "batch_loss": 46942.4921875, + "innovation_rms": 14880.32028520808, + "instruction_rms": 14880.320148654637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 380769280.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.367298448530919e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 368989.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1792.58837890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 439.2528076171875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8527116.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 453084.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28438.890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25409.94140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1605.368408203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 368989.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1792.58984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 439.2528076171875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8527116.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14880.320435427042, + "step": 140, + "teaching_rms": 14880.32028520808, + "traffic_rms": 2.9213298661258715 + }, + { + "batch_loss": 28753.6875, + "innovation_rms": 12001.57551193733, + "instruction_rms": 12001.57532106304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 587518208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9305684717364314e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 331238.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1801.4228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384.9183044433594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 455644.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25595.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25436.056640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1666.0244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 331238.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1801.42431640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384.9183044433594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12001.575722879803, + "step": 141, + "teaching_rms": 12001.57551193733, + "traffic_rms": 3.1054520062815887 + }, + { + "batch_loss": 16195.46875, + "innovation_rms": 109775.9651925522, + "instruction_rms": 109775.96517218684, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 805671872.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.537511536601858e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 297258.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809.35595703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 340.9581604003906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8589966.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 457947.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23035.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25459.560546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1663.9765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 297258.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809.357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 340.95819091796875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8589966.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 109775.96521511381, + "step": 142, + "teaching_rms": 109775.9651925522, + "traffic_rms": 3.0696885278088266 + }, + { + "batch_loss": 7430.72607421875, + "innovation_rms": 93113.69734236189, + "instruction_rms": 93113.69732184726, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010178176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1837603829416722e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266673.53125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1816.4776611328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303.6713562011719, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8616633.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 460020.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20731.94921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25480.712890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1566.0447998046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266673.53125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1816.4791259765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303.67138671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8616633.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 93113.69736510878, + "step": 143, + "teaching_rms": 93113.69734236189, + "traffic_rms": 2.8383274457601986 + }, + { + "batch_loss": 4553.58544921875, + "innovation_rms": 33081.90680798647, + "instruction_rms": 33081.906765117565, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1177340160.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.865384256686575e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239144.515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1822.8690185546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271.5569152832031, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8640547.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 461886.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18856.265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25499.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1457.2144775390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239144.515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1822.8704833984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271.55694580078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8640547.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33081.906855546156, + "step": 144, + "teaching_rms": 33081.90680798647, + "traffic_rms": 2.445925048436639 + }, + { + "batch_loss": 1524.624267578125, + "innovation_rms": 22796.16289090778, + "instruction_rms": 22796.162852018784, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1287683712.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5788458090276848e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214366.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1828.60302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 243.92755126953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8661984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 463565.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21594.22265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25516.884765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1332.101318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214366.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1828.6044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 243.92758178710938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8661984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22796.162934073775, + "step": 145, + "teaching_rms": 22796.16289090778, + "traffic_rms": 1.934041507024765 + }, + { + "batch_loss": 195.80589294433594, + "innovation_rms": 25777.57112942025, + "instruction_rms": 25777.571112710106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1327911552.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.320961294095614e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192063.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833.745361328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.53981018066406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681190.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 465077.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22656.904296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25532.3046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1198.89111328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192063.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833.746826171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.5398406982422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681190.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 25777.571147947478, + "step": 146, + "teaching_rms": 25777.57112942025, + "traffic_rms": 1.3478437539394923 + }, + { + "batch_loss": 159.88479614257812, + "innovation_rms": 9945457.076168535, + "instruction_rms": 9945457.076168522, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1291916672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0888651646860526e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171988.734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1838.3551025390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.58865356445312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698389.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 466437.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22142.82421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25546.18359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1079.001953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171988.734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1838.3565673828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.5886688232422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698389.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9945457.07616855, + "step": 147, + "teaching_rms": 9945457.076168535, + "traffic_rms": 0.732810979297577 + }, + { + "batch_loss": 142.257080078125, + "innovation_rms": 229021.32828031149, + "instruction_rms": 229021.32828029245, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1180918528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8799786262272147e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153920.015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1842.4854736328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.8306427001953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713781.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 467661.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20243.107421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25558.673828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 971.1017456054688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153920.015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1842.4869384765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.83065795898438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713781.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 229021.32828033034, + "step": 148, + "teaching_rms": 229021.32828031149, + "traffic_rms": 0.13183835471103542 + }, + { + "batch_loss": 510.8158874511719, + "innovation_rms": 130277294.75612995, + "instruction_rms": 130277294.75612995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1062830848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6919807196240282e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137656.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.184326171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.04664611816406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727547.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 468763.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18218.796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25569.916015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 873.9915771484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137656.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.185791015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.04666137695312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727547.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 130277294.75612995, + "step": 149, + "teaching_rms": 130277294.75612995, + "traffic_rms": 0.007074159689247733 + }, + { + "batch_loss": 20248.529296875, + "innovation_rms": 36049338306.00162, + "instruction_rms": 36049338306.00162, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 956613504.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.522782592686044e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123018.21875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.494873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.1591796875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739849.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 469755.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16396.916015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25580.033203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1367.1351318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123018.21875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.496337890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.15931701660156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739849.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 36049338306.00162, + "step": 150, + "teaching_rms": 36049338306.00162, + "traffic_rms": 0.29154960346644127 + }, + { + "batch_loss": 374.1198425292969, + "innovation_rms": 582427223.6219846, + "instruction_rms": 582427223.6219846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 860953152.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3705043004320907e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109842.40625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.455810546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.14675903320312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750833.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 470647.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14757.2236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25589.138671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230.421630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109842.40625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.457275390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.14691162109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750833.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 582427223.6219846, + "step": 151, + "teaching_rms": 582427223.6219846, + "traffic_rms": 0.0030013731878210308 + }, + { + "batch_loss": 11800.4111328125, + "innovation_rms": 5926944.797722141, + "instruction_rms": 5926944.797722137, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 774913088.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2334539143693468e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97983.078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.1021728515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.33750915527344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760631.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 471450.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13281.5009765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25597.333984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1107.37939453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97983.078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.1036376953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.33766174316406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760631.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5926944.797722138, + "step": 152, + "teaching_rms": 5926944.797722141, + "traffic_rms": 0.15938694507510712 + }, + { + "batch_loss": 853.9813842773438, + "innovation_rms": 3740.20667231608, + "instruction_rms": 3740.206853122594, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 697462400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1101085009421795e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87308.703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.46533203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.7033233642578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769362.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 472173.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11953.3505859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25604.708984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1029.9757080078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87308.703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.70347595214844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769362.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3740.206855394006, + "step": 153, + "teaching_rms": 3740.20667231608, + "traffic_rms": 0.13108184764498002 + }, + { + "batch_loss": 672.5816650390625, + "innovation_rms": 1997.540150362273, + "instruction_rms": 1997.5401425960727, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 627751168.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.99097628857729e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77700.890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.5736083984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.93087768554688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777132.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 472824.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10758.015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25611.34765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1315.9783935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77700.890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.5750732421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.9310302734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777132.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1997.540146139994, + "step": 154, + "teaching_rms": 1997.540150362273, + "traffic_rms": 0.11945034556914633 + }, + { + "batch_loss": 417.1689147949219, + "innovation_rms": 1720.1181991654296, + "instruction_rms": 1720.118192269717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 565003008.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.991878659719561e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69053.0859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.4525146484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.83416748046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784037.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 473409.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9682.2138671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25617.322265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1513.86181640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69053.0859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.4539794921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.83432006835938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784037.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1720.1181957883275, + "step": 155, + "teaching_rms": 1720.1181991654296, + "traffic_rms": 0.11017883574187924 + }, + { + "batch_loss": 185.9398956298828, + "innovation_rms": 1312.0361824316456, + "instruction_rms": 1312.0361095741766, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 508520704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.092690903698768e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61269.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.1248779296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.24578094482422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790164.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 473936.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713.9921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25622.69921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1629.765869140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61269.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.1263427734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.24593353271484, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790164.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1312.0361135633252, + "step": 156, + "teaching_rms": 1312.0361824316456, + "traffic_rms": 0.10262149527921566 + }, + { + "batch_loss": 59.91779327392578, + "innovation_rms": 16001.35647823779, + "instruction_rms": 16001.356477865094, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 457677792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.28342175835331e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54263.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.611328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.96595764160156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795590.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 474410.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7842.5927734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25627.5390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1668.3740234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54263.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.61279296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.96610260009766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795590.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16001.356478150627, + "step": 157, + "teaching_rms": 16001.35647823779, + "traffic_rms": 0.09559555073251423 + }, + { + "batch_loss": 97.5270767211914, + "innovation_rms": 12437.105777768933, + "instruction_rms": 12437.105777443967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 411917344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.555079417591235e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47957.51953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.9305419921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.96207427978516, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800386.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 474837.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7058.33349609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25631.89453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1636.314453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47957.51953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.9320068359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.96221160888672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800386.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12437.105777768007, + "step": 158, + "teaching_rms": 12437.105777768933, + "traffic_rms": 0.08973776951162989 + }, + { + "batch_loss": 108.66709899902344, + "innovation_rms": 12425.852663317302, + "instruction_rms": 12425.85266306827, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 370733760.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.899571475832111e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42281.7265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.09912109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.05758666992188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804614.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 475221.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6352.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25635.814453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1547.53662109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42281.7265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.1005859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.05772399902344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804614.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12425.852663351983, + "step": 159, + "teaching_rms": 12425.852663317302, + "traffic_rms": 0.08387160331231051 + }, + { + "batch_loss": 28.771549224853516, + "innovation_rms": 4743.349480317385, + "instruction_rms": 4743.349773190418, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 333666048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.309614218297737e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37173.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.1322021484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.14265441894531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808331.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 475567.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5717.25, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25639.341796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1449.739990234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37173.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.1336669921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.14279174804688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808331.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4743.349773712865, + "step": 160, + "teaching_rms": 4743.349480317385, + "traffic_rms": 0.07039108685593892 + }, + { + "batch_loss": 49.8839225769043, + "innovation_rms": 5174.947829060944, + "instruction_rms": 5174.947828824574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 300302752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.778652906419126e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32574.951171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.0433349609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.11841583251953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811588.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 475878.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5145.52490234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25642.517578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1349.797607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32574.951171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.0447998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.1185531616211, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811588.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5174.947829111238, + "step": 161, + "teaching_rms": 5174.947829060944, + "traffic_rms": 0.05446490643448948 + }, + { + "batch_loss": 137.26068115234375, + "innovation_rms": 4342.474463375717, + "instruction_rms": 4342.474463230649, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 270274528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.300787560801632e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28436.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8446044921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.89588165283203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476158.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4630.97216796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25645.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1248.7637939453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28436.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8460693359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.89601135253906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4342.474463403391, + "step": 162, + "teaching_rms": 4342.474463375717, + "traffic_rms": 0.03858703743423635 + }, + { + "batch_loss": 5.444925785064697, + "innovation_rms": 8165.902048254391, + "instruction_rms": 8165.902048218558, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 243248400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.870708914672632e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24711.22265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.547119140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.39495086669922, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816903.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476411.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4167.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25647.947265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1147.6409912109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24711.22265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.548583984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.395084381103516, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816903.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8165.902048255262, + "step": 163, + "teaching_rms": 8165.902048254391, + "traffic_rms": 0.024483450354417305 + }, + { + "batch_loss": 3.961790084838867, + "innovation_rms": 1952.2187359412437, + "instruction_rms": 1952.2188249660958, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 218924272.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4836379407419965e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21358.41015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.16064453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.54352951049805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819039.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476637.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3751.08740234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25650.26171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1047.2542724609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21358.41015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.162109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.54366683959961, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819039.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1952.2188250090103, + "step": 164, + "teaching_rms": 1952.2187359412437, + "traffic_rms": 0.012944329831814956 + }, + { + "batch_loss": 3.811180591583252, + "innovation_rms": 0.265417077280466, + "instruction_rms": 0.26531595657338863, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 197031840.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1352741741555876e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18340.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.694091796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.27672576904297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820873.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476842.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3375.978515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25652.345703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 942.52880859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18340.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.695556640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.27686309814453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820873.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26531595657338863, + "step": 165, + "teaching_rms": 0.265417077280466, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.744426727294922, + "innovation_rms": 0.2871283620189236, + "instruction_rms": 0.28708262912720606, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 177328656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8217467842278195e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15624.5107421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.155517578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.536128997802734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822435.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477025.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3038.380615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25654.220703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 848.27587890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15624.5107421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.156982421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.5362663269043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822435.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28708262912720606, + "step": 166, + "teaching_rms": 0.2871283620189236, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.9801549911499023, + "innovation_rms": 0.2541226041526663, + "instruction_rms": 0.2540026937672959, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 159595792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.539571968366084e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13179.81640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.552001953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.269168853759766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823753.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477191.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2734.54248046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25655.908203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 763.4482421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13179.81640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.553466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.26930236816406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823753.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2540026937672959, + "step": 167, + "teaching_rms": 0.2541226041526663, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.7612504959106445, + "innovation_rms": 0.2875131544099313, + "instruction_rms": 0.28745403210754383, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 143636208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.285614744041685e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10979.458984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.89013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.42851638793945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824851.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477340.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2461.088134765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25657.427734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 687.1033935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10979.458984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.8916015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.428653717041016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824851.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28745403210754383, + "step": 168, + "teaching_rms": 0.2875131544099313, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.6806769371032715, + "innovation_rms": 0.3060041205106952, + "instruction_rms": 0.30598188373525176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 129272584.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.057053214661935e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9245.9404296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.17578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.971588134765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825751.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477473.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2214.979248046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25658.794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 618.39306640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9245.9404296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.17724609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.97172737121582, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825751.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30598188373525176, + "step": 169, + "teaching_rms": 0.3060041205106952, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.676483154296875, + "innovation_rms": 0.2625952381519611, + "instruction_rms": 0.2624638352844185, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 116345328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8513479206835323e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8291.470703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.4140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.860042572021484, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826473.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477594.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1993.4813232421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25660.025390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 556.5537719726562, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8291.470703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.41552734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.86018180847168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826473.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2624638352844185, + "step": 170, + "teaching_rms": 0.2625952381519611, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.4520416259765625, + "innovation_rms": 0.28537347654501183, + "instruction_rms": 0.2852918655692243, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104710792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6662131148712837e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7432.365234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.60986328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.05937385559082, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827034.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477703.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1794.1331787109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25661.1328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 500.89837646484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7432.365234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.611328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.059513092041016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827034.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2852918655692243, + "step": 171, + "teaching_rms": 0.28537347654501183, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.229200601577759, + "innovation_rms": 0.276863496340481, + "instruction_rms": 0.2767469030771587, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 94239712.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.499591748408574e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6659.09619140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.767333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.53852081298828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827451.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477800.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1614.7198486328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25662.12890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 450.80853271484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6659.09619140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.768798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.538660049438477, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827451.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2767469030771587, + "step": 172, + "teaching_rms": 0.276863496340481, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.019123077392578, + "innovation_rms": 0.2567116537650224, + "instruction_rms": 0.2565338532666274, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 84815744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3496325185921352e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5963.087890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.8902587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.269527435302734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827738.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477888.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1453.247802734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25663.025390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 405.7276611328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5963.087890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.8917236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.269668579101562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827738.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2565338532666274, + "step": 173, + "teaching_rms": 0.2567116537650224, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.127912998199463, + "innovation_rms": 0.2642271291224182, + "instruction_rms": 0.2641103934156552, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 76334168.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2146692529890263e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5336.6201171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.982177734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.227231979370117, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827908.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477967.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1307.9229736328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25663.83203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 365.1548767089844, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5336.6201171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.983642578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.227371215820312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827908.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2641103934156552, + "step": 174, + "teaching_rms": 0.2642271291224182, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.1852471828460693, + "innovation_rms": 0.2978679855502292, + "instruction_rms": 0.2978165663744342, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 68700752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0932023551779144e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4772.74609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.046142578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.38898277282715, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827973.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478038.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1177.130615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25664.55859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 328.6393737792969, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4772.74609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.047607421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.389123916625977, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827973.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2978165663744342, + "step": 175, + "teaching_rms": 0.2978679855502292, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.0689573287963867, + "innovation_rms": 0.27477864881462377, + "instruction_rms": 0.2747028713300193, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 61830676.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.838821059162276e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4265.21142578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.734395027160645, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827943.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478102.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1059.41748046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25665.212890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 295.7754211425781, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4265.21142578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.08642578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.734536170959473, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827943.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2747028713300193, + "step": 176, + "teaching_rms": 0.27477864881462377, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.4559617042541504, + "innovation_rms": 0.28780160467903726, + "instruction_rms": 0.2877768807582832, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 55647608.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.854938884526572e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3808.387939453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.1011962890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.245118141174316, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827828.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478160.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 953.4757080078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25665.80078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.1978759765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3808.387939453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.1026611328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.245260238647461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827828.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2877768807582832, + "step": 177, + "teaching_rms": 0.28780160467903726, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.0356574058532715, + "innovation_rms": 0.279009028345581, + "instruction_rms": 0.2789587593754558, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 50082848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.969445133512868e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3397.208740234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0970458984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.90463638305664, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827636.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478212.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858.1281127929688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25666.330078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239.5780792236328, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3397.208740234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0985107421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.904778480529785, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827636.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27895875937545755, + "step": 178, + "teaching_rms": 0.279009028345581, + "traffic_rms": 3.1193397810880315e-08 + }, + { + "batch_loss": 2.94931960105896, + "innovation_rms": 3206.2685510452966, + "instruction_rms": 3206.2685505553663, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 45082540.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.172500551442104e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3027.113525390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0745849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.698083877563477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827375.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478259.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 772.3153076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25666.806640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 215.62026977539062, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3027.113525390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0760498046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.698226928710938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827375.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3206.2685510584815, + "step": 179, + "teaching_rms": 3206.2685510452966, + "traffic_rms": 0.05680004858165525 + }, + { + "batch_loss": 2.8555386066436768, + "innovation_rms": 10899.010887625855, + "instruction_rms": 10899.010887361406, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 40581720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.455250633736847e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55532.4609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.03564453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.612079620361328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827052.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478301.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 695.083740234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25667.236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.41537475585938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55532.4609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.037109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.612222671508789, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827052.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10899.010887597187, + "step": 180, + "teaching_rms": 10899.010887625855, + "traffic_rms": 0.07169058495221262 + }, + { + "batch_loss": 53.69892883300781, + "innovation_rms": 1632683.8899226976, + "instruction_rms": 1632683.8899226934, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36528344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.809725295485256e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49978.671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9818115234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.634580612182617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826673.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478339.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 625.5753784179688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25667.623046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 440.19635009765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49978.671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9832763671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.634723663330078, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826673.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1632683.8899226945, + "step": 181, + "teaching_rms": 1632683.8899226976, + "traffic_rms": 0.057424240030186625 + }, + { + "batch_loss": 2.85539174079895, + "innovation_rms": 1437.0915126469017, + "instruction_rms": 1437.0915047528385, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32879792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.2287526284977766e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44979.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9146728515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.754745006561279, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826243.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478373.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 563.017822265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25667.970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 754.9651489257812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44979.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9161376953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.75488805770874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826243.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1437.0915062697325, + "step": 182, + "teaching_rms": 1437.0915126469017, + "traffic_rms": 0.0660287638007174 + }, + { + "batch_loss": 923.4841918945312, + "innovation_rms": 3788100.555174392, + "instruction_rms": 3788100.5551743917, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29592286.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.705877468727214e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40480.28125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.83544921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.476913452148438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478403.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 506.7160339355469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25668.283203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 679.4686279296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40480.28125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.8369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.476882934570312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3788100.555174392, + "step": 183, + "teaching_rms": 3788100.555174392, + "traffic_rms": 0.03765189622279445 + }, + { + "batch_loss": 404.11505126953125, + "innovation_rms": 501391.4005809951, + "instruction_rms": 501391.4005809878, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26635186.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.235289584415539e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36430.34765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.7454833984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.724916458129883, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825252.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478431.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 456.04443359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25668.564453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 611.521728515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36430.34765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.7469482421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.724885940551758, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825252.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 501391.40058099496, + "step": 184, + "teaching_rms": 501391.4005809951, + "traffic_rms": 0.08499668843558371 + }, + { + "batch_loss": 355.9176940917969, + "innovation_rms": 891291.8584684258, + "instruction_rms": 891291.8584684238, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 23971398.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8117604541752934e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32785.04296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.645751953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.949853897094727, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824700.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478456.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410.4399719238281, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25668.818359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 550.3695678710938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32785.04296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.647216796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.949827194213867, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824700.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 891291.8584684242, + "step": 185, + "teaching_rms": 891291.8584684258, + "traffic_rms": 0.07881819347698286 + }, + { + "batch_loss": 334.0430908203125, + "innovation_rms": 1209558.4590948278, + "instruction_rms": 1209558.4590948175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21574060.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.430584511836979e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29503.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.5372314453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.34711265563965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824115.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478478.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 369.3959655761719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 517.6212158203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29503.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.5386962890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.34708595275879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824115.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1209558.4590948238, + "step": 186, + "teaching_rms": 1209558.4590948278, + "traffic_rms": 0.06519848096930188 + }, + { + "batch_loss": 224.1370849609375, + "innovation_rms": 1617794.919163045, + "instruction_rms": 1617794.9191630513, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19416592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0875258888545894e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26550.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.4208984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.94455623626709, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823500.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478498.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 332.45635986328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.251953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 558.976318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26550.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.42236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.944528579711914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823500.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1617794.919163052, + "step": 187, + "teaching_rms": 1617794.919163045, + "traffic_rms": 0.0463195458421253 + }, + { + "batch_loss": 220.3536834716797, + "innovation_rms": 874731.0354493351, + "instruction_rms": 874731.0354493349, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17475652.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7787732312496538e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23892.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.29736328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.669665336608887, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822858.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478516.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 299.2107238769531, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 540.0653686523438, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23892.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.298828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.669637680053711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822858.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 874731.0354493352, + "step": 188, + "teaching_rms": 874731.0354493351, + "traffic_rms": 0.026226356132063647 + }, + { + "batch_loss": 2.8601067066192627, + "innovation_rms": 9043.650425852305, + "instruction_rms": 9043.65042584984, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15730073.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5008958909448192e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21499.779296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.16748046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.304486274719238, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478532.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 269.2896423339844, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.603515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 486.4154968261719, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21499.779296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.1689453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.304458618164062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9043.650425859065, + "step": 189, + "teaching_rms": 9043.650425852305, + "traffic_rms": 0.012991948438177 + }, + { + "batch_loss": 1286.6353759765625, + "innovation_rms": 121198932.52707139, + "instruction_rms": 121198932.52707139, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14162339.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2508063018503373e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19346.185546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.0318603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.075702667236328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8821505.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478547.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.3606719970703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.75390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 437.77392578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19346.185546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.0333251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.075675010681152, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8821505.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 121198932.52707139, + "step": 190, + "teaching_rms": 121198932.52707139, + "traffic_rms": 0.02878346767642578 + }, + { + "batch_loss": 2.9269120693206787, + "innovation_rms": 0.29738323338221917, + "instruction_rms": 0.29734029032163867, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12746105.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.025725620125696e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17407.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.89111328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.969685554504395, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820798.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478560.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 218.12460327148438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.888671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 393.99652099609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17407.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.892578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.969658851623535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820798.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29751377905242204, + "step": 191, + "teaching_rms": 0.29738323338221917, + "traffic_rms": 0.010158763581116153 + }, + { + "batch_loss": 2.598923683166504, + "innovation_rms": 352878.727046624, + "instruction_rms": 352878.72704661876, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11485154.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8231529722137805e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29846.158203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.7457275390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.974170684814453, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820074.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478572.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.3121337890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.009765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 360.0180969238281, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29846.158203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.7471923828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.974143981933594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820074.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 352878.7270466288, + "step": 192, + "teaching_rms": 352878.727046624, + "traffic_rms": 0.08392932259073368 + }, + { + "batch_loss": 932.2616577148438, + "innovation_rms": 112207387.69580767, + "instruction_rms": 112207387.69580767, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10339830.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.640837623452795e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26861.82421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.5960693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.078118324279785, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819334.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478583.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.41970825195312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.119140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 644.5219116210938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26861.82421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.5975341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.078091621398926, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819334.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 112207387.69580767, + "step": 193, + "teaching_rms": 112207387.69580767, + "traffic_rms": 0.013491933412467738 + }, + { + "batch_loss": 29.73762321472168, + "innovation_rms": 142702.85970973683, + "instruction_rms": 142702.85970973506, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9306102.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4767538954672538e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24175.07421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.442626953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.34222984313965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8818580.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478593.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.2609100341797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.216796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 542.1224975585938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24175.07421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.444091796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.34255027770996, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8818580.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 142702.85970973465, + "step": 194, + "teaching_rms": 142702.85970973683, + "traffic_rms": 0.007928532531607906 + }, + { + "batch_loss": 147.31788635253906, + "innovation_rms": 217291.06349472623, + "instruction_rms": 217291.0634947254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8376691.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3290784887406592e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21756.423828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.285888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.536293029785156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817813.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478601.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 149.1409454345703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.3046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487.9102478027344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21756.423828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.287353515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.536338806152344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817813.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 217291.0634947265, + "step": 195, + "teaching_rms": 217291.06349472623, + "traffic_rms": 0.022275275921000467 + }, + { + "batch_loss": 98.02535247802734, + "innovation_rms": 87151.23327453088, + "instruction_rms": 87151.23327452905, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7540019.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1961706398665933e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19579.509765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.1260986328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.383174896240234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817034.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478609.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128.8003692626953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.384765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 439.11920166015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19579.509765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.1275634765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.383087158203125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817034.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 87151.23327453154, + "step": 196, + "teaching_rms": 87151.23327453088, + "traffic_rms": 0.02093597297506642 + }, + { + "batch_loss": 110.49617767333984, + "innovation_rms": 9651.219023627074, + "instruction_rms": 9651.219023603622, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6787064.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0765535672899994e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17620.150390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.9635009765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.24050521850586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816245.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478616.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.92032623291016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.45703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 395.207275390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17620.150390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.9649658203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.2403678894043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816245.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9651.219023623473, + "step": 197, + "teaching_rms": 9651.219023627074, + "traffic_rms": 0.01986103000747594 + }, + { + "batch_loss": 139.51123046875, + "innovation_rms": 17968.384378315954, + "instruction_rms": 17968.384378292307, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6109267.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.688982363308032e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15856.587890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.7984619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.284793853759766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8815447.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478622.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.51786804199219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.521484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.6865234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15856.587890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.7999267578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.284656524658203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8815447.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17968.384378301376, + "step": 198, + "teaching_rms": 17968.384378315954, + "traffic_rms": 0.018495063933818157 + }, + { + "batch_loss": 309.222900390625, + "innovation_rms": 1516852.2388912716, + "instruction_rms": 1516852.2388912714, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5499099.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.720084126977229e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14269.26171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.6312255859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.9809627532959, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478628.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.46914672851562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.580078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 320.11785888671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14269.26171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.6326904296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.980825424194336, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814640.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1516852.2388912714, + "step": 199, + "teaching_rms": 1516852.2388912716, + "traffic_rms": 0.01711390573592764 + }, + { + "batch_loss": 93.79814910888672, + "innovation_rms": 44566.17938654317, + "instruction_rms": 44566.17938653945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4949894.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.84807605787689e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12840.537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.4620361328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.999473571777344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813826.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478633.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.3681869506836, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.6328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288.1060791015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12840.537109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.4635009765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.999500274658203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813826.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 44566.179386541866, + "step": 200, + "teaching_rms": 44566.17938654317, + "traffic_rms": 0.015386626809295773 + }, + { + "batch_loss": 6.241382122039795, + "innovation_rms": 15861.076889584181, + "instruction_rms": 15861.076889578206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4455546.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.063268280290509e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11554.57421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.198814392089844, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813005.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478638.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.16688537597656, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.6796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 259.29547119140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11554.57421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.29248046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.198843002319336, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813005.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15861.076889584154, + "step": 201, + "teaching_rms": 15861.076889584181, + "traffic_rms": 0.01376293302916301 + }, + { + "batch_loss": 2.875265121459961, + "innovation_rms": 1980.7366439625214, + "instruction_rms": 1980.7366439359062, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4010560.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.356941581110477e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10397.10546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.118408203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.57800579071045, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8812178.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478642.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.1427230834961, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.72265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.3659210205078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10397.10546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.119873046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.578034400939941, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8812178.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1980.7366439730527, + "step": 202, + "teaching_rms": 1980.7366439625214, + "traffic_rms": 0.012130779320862147 + }, + { + "batch_loss": 2.7996561527252197, + "innovation_rms": 2143.531357866749, + "instruction_rms": 2143.5313578481628, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3610000.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.721247337100083e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9355.2919921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.9443359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.119132041931152, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811346.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478646.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130.58822631835938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.759765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210.02932739257812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9355.2919921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.94580078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.119160652160645, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811346.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2143.531357873992, + "step": 203, + "teaching_rms": 2143.531357866749, + "traffic_rms": 0.01052291113050126 + }, + { + "batch_loss": 2.723015308380127, + "innovation_rms": 4644.328874188232, + "instruction_rms": 4644.328874183763, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3249423.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.149122474541056e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8417.576171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.7689208984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.806015014648438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8810509.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478649.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.09407043457031, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 189.02638244628906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8417.576171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.7703857421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.80604362487793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8810509.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4644.328874192497, + "step": 204, + "teaching_rms": 4644.328874188232, + "traffic_rms": 0.009006375942675771 + }, + { + "batch_loss": 2.8032076358795166, + "innovation_rms": 31318.218841325594, + "instruction_rms": 31318.218841325328, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2924843.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.634210012338586e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8415.279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.5924072265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.624091148376465, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8809667.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478652.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.77883911132812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.826171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.1237335205078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8415.279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.5938720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.624119758605957, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8809667.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31318.21884132626, + "step": 205, + "teaching_rms": 31318.218841325594, + "traffic_rms": 0.0076401103106344456 + }, + { + "batch_loss": 2.8712570667266846, + "innovation_rms": 0.2764500082549549, + "instruction_rms": 0.2764700626176387, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2632359.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1707890540544e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8454.7177734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.414794921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.560254096984863, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808822.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478655.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.2009506225586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.853515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.11135864257812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8454.7177734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.416259765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.560282707214355, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808822.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2764700626176387, + "step": 206, + "teaching_rms": 0.2764500082549549, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.8995115756988525, + "innovation_rms": 0.3022319427536631, + "instruction_rms": 0.3022781422600584, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2369123.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.753709933900595e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490.1279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.2362060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.602704048156738, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807973.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478657.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.78085327148438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.87890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.8002166748047, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490.1279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.2376708984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.60273265838623, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807973.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3022781422600584, + "step": 207, + "teaching_rms": 0.3022319427536631, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.719717502593994, + "innovation_rms": 0.25785018801185927, + "instruction_rms": 0.2578412468404832, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2132211.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3783390264098816e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8521.912109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.0567626953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.740824222564697, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807121.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478659.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.40276336669922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.900390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.02019500732422, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8521.912109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.0582275390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.740852355957031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807121.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2578412468404832, + "step": 208, + "teaching_rms": 0.25785018801185927, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.8080978393554688, + "innovation_rms": 0.29331971850466637, + "instruction_rms": 0.29334658818648046, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1918989.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.040505059344384e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8550.4326171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8765869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.965054512023926, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8806266.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478661.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.96248626708984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.919921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.61817169189453, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8550.4326171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8780517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.96508264541626, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8806266.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29334658818648046, + "step": 209, + "teaching_rms": 0.29331971850466637, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.8100006580352783, + "innovation_rms": 0.31482099660098894, + "instruction_rms": 0.3148579413911109, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1727090.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.736454446035763e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8576.015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.69580078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.266792297363281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8805408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478663.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.3662338256836, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.45635223388672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8576.015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.697265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.266820430755615, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8805408.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3148579413911109, + "step": 210, + "teaching_rms": 0.31482099660098894, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.787254571914673, + "innovation_rms": 0.312496978205725, + "instruction_rms": 0.3125383768151723, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1554381.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4628089799573504e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8598.955078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.514404296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.638293743133545, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804548.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478665.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.52960968017578, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.41071319580078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8598.955078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.515869140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.638321876525879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804548.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3125383768151723, + "step": 211, + "teaching_rms": 0.312496978205725, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.6707780361175537, + "innovation_rms": 0.272209829233751, + "instruction_rms": 0.2722155449909677, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1398943.625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2165280390119424e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8619.5146484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.3323974609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0725884437561035, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8803686.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478666.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.37664794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.96875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.36963653564453, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8619.5146484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.3338623046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0726165771484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8803686.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2722155449909677, + "step": 212, + "teaching_rms": 0.272209829233751, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.786745309829712, + "innovation_rms": 0.30849719063652636, + "instruction_rms": 0.3085369924984441, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1259049.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9948752995352576e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8637.931640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.14990234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.563403129577637, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8802822.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478667.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.83898162841797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.982421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.23267364501953, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8637.931640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.1513671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.563431262969971, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8802822.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3085369924984441, + "step": 213, + "teaching_rms": 0.30849719063652636, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.490892171859741, + "innovation_rms": 0.27242623375465413, + "instruction_rms": 0.2723966703730545, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1133144.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7953878125314048e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8654.419921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.9669189453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.105090618133545, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801957.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478668.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.85508346557617, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.994140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.9094009399414, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8654.419921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.9683837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.105118751525879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801957.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2723966703730545, + "step": 214, + "teaching_rms": 0.27242623375465413, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5046255588531494, + "innovation_rms": 0.240116748511423, + "instruction_rms": 0.24006809427879872, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1019829.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6158490527531008e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8669.1728515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.7835693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.69256854057312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801090.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478670.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.36957550048828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.00390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.318458557128906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8669.1728515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.7850341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.692596435546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801090.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24006809427879872, + "step": 215, + "teaching_rms": 0.240116748511423, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.613438606262207, + "innovation_rms": 0.2947589929951547, + "instruction_rms": 0.2947654671844913, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 917846.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.454264115265536e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682.3642578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.599853515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3212616443634033, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800222.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478670.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.332618713378906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.013671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.38661193847656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682.3642578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.601318359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.321289300918579, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800222.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2947654671844913, + "step": 216, + "teaching_rms": 0.2947589929951547, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5097427368164062, + "innovation_rms": 0.28831718111188115, + "instruction_rms": 0.28828831242971514, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 826062.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3088377037389824e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.1494140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.4158935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9870522022247314, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8799353.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478671.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.69935607910156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.021484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.047950744628906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.1494140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.4173583984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9870798587799072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8799353.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28828831242971514, + "step": 217, + "teaching_rms": 0.28831718111188115, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.620239019393921, + "innovation_rms": 0.3104054756979925, + "instruction_rms": 0.3104137440924104, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 743456.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1779539441025024e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.669921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.2315673828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6862337589263916, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8798483.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478672.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.429420471191406, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.029296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.2431526184082, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.669921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.2330322265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6862616539001465, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8798483.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3104137440924104, + "step": 218, + "teaching_rms": 0.3104054756979925, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.599306344985962, + "innovation_rms": 0.3015312405059806, + "instruction_rms": 0.30154038821056556, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 669110.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0601585604296704e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.05078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.0469970703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4154703617095947, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8797612.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478673.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.48647689819336, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.037109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.9188346862793, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.05078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.0484619140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4154982566833496, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8797612.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30154038821056556, + "step": 219, + "teaching_rms": 0.3015312405059806, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.47062087059021, + "innovation_rms": 0.25058714309037916, + "instruction_rms": 0.25052363041756415, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 602199.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9541426614370304.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.40625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.8621826171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1717591285705566, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8796740.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478673.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.837827682495117, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.04296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.02695083618164, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.40625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.8636474609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1717870235443115, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8796740.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25052363041756415, + "step": 220, + "teaching_rms": 0.25058714309037916, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.531863212585449, + "innovation_rms": 0.30875829962950296, + "instruction_rms": 0.30875452584560603, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 541979.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8587284060307456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.83984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.67724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.952397346496582, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795867.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478674.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.454044342041016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.048828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.524255752563477, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.83984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.6787109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.952425241470337, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795867.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30875452584560603, + "step": 221, + "teaching_rms": 0.30875829962950296, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5695884227752686, + "innovation_rms": 0.27945092925617243, + "instruction_rms": 0.27943629875873455, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 487781.4375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7728555815337984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.4423828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.4920654296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8514926433563232, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794993.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478674.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.308639526367188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.371829986572266, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.4423828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.4935302734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8514940738677979, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794993.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27943629875873455, + "step": 222, + "teaching_rms": 0.27945092925617243, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4404056072235107, + "innovation_rms": 0.2750626279261716, + "instruction_rms": 0.27502620557083274, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 439003.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6955700072742912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.3067626953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8530925512313843, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794119.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478675.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.377775192260742, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.05859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.53464698791504, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.3082275390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8530939817428589, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794119.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27502620557083274, + "step": 223, + "teaching_rms": 0.2750626279261716, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5614099502563477, + "innovation_rms": 0.26358463843060204, + "instruction_rms": 0.26355341463224524, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 395102.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6260130119155712.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.478515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.121337890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8545138835906982, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8793244.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478675.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.639997482299805, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.981182098388672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.478515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.122802734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8545154333114624, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8793244.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26355341463224524, + "step": 224, + "teaching_rms": 0.26358463843060204, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.579437494277954, + "innovation_rms": 0.3108100442774422, + "instruction_rms": 0.31083092491413566, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 355592.65625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5634117160927232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.0546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.935791015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8557746410369873, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8792369.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478676.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.075997352600098, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.06640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.683063507080078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.0546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.937255859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855776071548462, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8792369.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.31083092491413566, + "step": 225, + "teaching_rms": 0.3108100442774422, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.408668041229248, + "innovation_rms": 0.2473100742409287, + "instruction_rms": 0.2472415061541871, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 320033.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5070705498521600.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.0859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.7501220703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568906784057617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8791493.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478676.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.668396949768066, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.614757537841797, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.0859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.7515869140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568921089172363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8791493.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2472415061541871, + "step": 226, + "teaching_rms": 0.2473100742409287, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4933738708496094, + "innovation_rms": 0.2925992066039426, + "instruction_rms": 0.2925895470351429, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 288030.03125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4563634680233984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.6259765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.5643310546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8578765392303467, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790617.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478676.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.401556968688965, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.07421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.753280639648438, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.6259765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.5657958984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8578780889511108, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790617.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2925895470351429, + "step": 227, + "teaching_rms": 0.2925992066039426, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4220948219299316, + "innovation_rms": 0.29562838644358086, + "instruction_rms": 0.295593358431202, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 259227.03125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4107271319584768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.724609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.37841796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8587453365325928, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8789741.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.261401176452637, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.07795238494873, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.724609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.3798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.858746886253357, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8789741.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.295593358431202, + "step": 228, + "teaching_rms": 0.29562838644358086, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.380352020263672, + "innovation_rms": 0.2618839761268197, + "instruction_rms": 0.26182863774098936, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 233304.328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3696544133939200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.42578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.1925048828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8595086336135864, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8788864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.235260963439941, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.570157051086426, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.42578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.1939697265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8595101833343506, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8788864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26182863774098936, + "step": 229, + "teaching_rms": 0.2618839761268197, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4585654735565186, + "innovation_rms": 0.29144988141841616, + "instruction_rms": 0.2914303489615634, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 209973.890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3326889720545280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.76953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.0064697265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8601770401000977, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787987.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.311734199523926, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.080078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.213141441345215, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.76953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.0079345703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8601784706115723, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787987.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2914303489615634, + "step": 230, + "teaching_rms": 0.29144988141841616, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4877355098724365, + "innovation_rms": 0.28743268844788356, + "instruction_rms": 0.2874216380348772, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 188976.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2994200748490752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.791015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.8204345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860759973526001, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787110.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.480560779571533, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.08203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.991827011108398, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.791015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.8218994140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8607614040374756, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787110.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2874216380348772, + "step": 231, + "teaching_rms": 0.28743268844788356, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.410398006439209, + "innovation_rms": 0.3091903203571299, + "instruction_rms": 0.3091840876191504, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 170078.84375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2694780593111040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.5224609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.63427734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861266016960144, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8786233.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.732504367828369, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.892643928527832, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.5224609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.6357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8612674474716187, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8786233.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3091840876191504, + "step": 232, + "teaching_rms": 0.3091903203571299, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.364511251449585, + "innovation_rms": 0.28030154143462943, + "instruction_rms": 0.28026665984351146, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 153070.953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2425302533799936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.9931640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.4481201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861702799797058, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8785356.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.059253692626953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.903379440307617, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.9931640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.4495849609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8617042303085327, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8785356.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28026665984351146, + "step": 233, + "teaching_rms": 0.28030154143462943, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3918378353118896, + "innovation_rms": 0.274034241004796, + "instruction_rms": 0.2739853198889661, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 137763.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2182772307263488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.2294921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.261962890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862077236175537, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784479.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4533281326293945, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.087890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.013041496276855, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.2294921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.263427734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8620786666870117, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784479.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2739853198889661, + "step": 234, + "teaching_rms": 0.274034241004796, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.341724395751953, + "innovation_rms": 0.2688285220655465, + "instruction_rms": 0.26875931132572245, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 123987.4765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1964495089958912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.25390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.07568359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623956441879272, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8783601.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.907995223999023, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.08984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.211737155914307, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.25390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.0771484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623970746994019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8783601.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26875931132572245, + "step": 235, + "teaching_rms": 0.2688285220655465, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.383183002471924, + "innovation_rms": 0.28960502818337375, + "instruction_rms": 0.289569602682948, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 111588.7265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1768045567541248.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.0888671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.889404296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8626636266708374, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8782723.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.417195796966553, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.49056339263916, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.0888671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.890869140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862665057182312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8782723.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.289569602682948, + "step": 236, + "teaching_rms": 0.28960502818337375, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3871207237243652, + "innovation_rms": 0.29276195018589, + "instruction_rms": 0.29273233356127637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 100429.8515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1591240957100032.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.751953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8628861904144287, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8781845.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.975476026535034, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.8415069580078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.751953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.70458984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8628876209259033, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8781845.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29273233356127637, + "step": 237, + "teaching_rms": 0.29276195018589, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2803874015808105, + "innovation_rms": 0.26279037099768104, + "instruction_rms": 0.26270405280205833, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 90386.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1432116847968256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.2607421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.516845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8630677461624146, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780967.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.577928304672241, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.2573561668396, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.2607421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.518310546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8630692958831787, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780967.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26270405280205833, + "step": 238, + "teaching_rms": 0.26279037099768104, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.362626314163208, + "innovation_rms": 0.2648427387627951, + "instruction_rms": 0.26476164119025986, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 81348.1796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1288905190014976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.630859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.33056640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8632125854492188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780089.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.220135450363159, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.7316203117370605, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.630859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.33203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863214135169983, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780089.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26476164119025986, + "step": 239, + "teaching_rms": 0.2648427387627951, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3330788612365723, + "innovation_rms": 0.2663034094427114, + "instruction_rms": 0.2662159426901059, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 73213.359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1160014697857024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.876953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.144287109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8633244037628174, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8779211.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8981218338012695, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.258458137512207, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.876953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.145751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863325834274292, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8779211.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2662159426901059, + "step": 240, + "teaching_rms": 0.2663034094427114, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.332080602645874, + "innovation_rms": 0.2821871271493446, + "instruction_rms": 0.28211482777303043, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 65892.0234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1044013201227776.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.009765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.9578857421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634064197540283, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778333.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.608309507369995, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8326122760772705, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.009765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.9593505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863407850265503, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778333.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28211482777303043, + "step": 241, + "teaching_rms": 0.2821871271493446, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3469736576080322, + "innovation_rms": 0.2735907047472714, + "instruction_rms": 0.27351144300281705, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 59302.8203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 939611874394112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.0419921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634614944458008, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777455.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3474783897399902, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4493510723114014, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.0419921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.77294921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634629249572754, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777455.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27351144300281705, + "step": 242, + "teaching_rms": 0.2735907047472714, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3143692016601562, + "innovation_rms": 0.2823324576365145, + "instruction_rms": 0.28226663148077835, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 53372.5390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 845650673532928.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.9833984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.5850830078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863492488861084, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776577.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1127305030822754, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1044158935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.9833984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.5865478515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634939193725586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776577.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28226663148077835, + "step": 243, + "teaching_rms": 0.2823324576365145, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3266451358795166, + "innovation_rms": 0.25587522192346895, + "instruction_rms": 0.25577527324758226, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 48035.28515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 761085586046976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.8427734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.398681640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8635016679763794, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775699.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.90145742893219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7939741611480713, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.8427734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.400146484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8635032176971436, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775699.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25577527324758226, + "step": 244, + "teaching_rms": 0.25587522192346895, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3292765617370605, + "innovation_rms": 0.2878560406695384, + "instruction_rms": 0.2877940532811175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43231.7578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 684977020731392.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.6279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.2122802734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634912967681885, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774821.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7113116979599, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5145766735076904, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.6279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.2137451171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634928464889526, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774821.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2877940532811175, + "step": 245, + "teaching_rms": 0.2878560406695384, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.306917190551758, + "innovation_rms": 0.2782939196186503, + "instruction_rms": 0.27821093053156254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 38908.58203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 616479338790912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.3466796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.02587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634634017944336, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773943.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5401804447174072, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2631189823150635, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.3466796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.02734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634649515151978, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773943.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27821093053156254, + "step": 246, + "teaching_rms": 0.2782939196186503, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31752347946167, + "innovation_rms": 0.2871232895858414, + "instruction_rms": 0.287053244609099, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35022.21875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 554831391490048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.005859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.8394775390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863419532775879, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773065.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3861624002456665, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.036807060241699, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.005859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.8409423828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634212017059326, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773065.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.287053244609099, + "step": 247, + "teaching_rms": 0.2871232895858414, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3311145305633545, + "innovation_rms": 0.27202173926514006, + "instruction_rms": 0.2719526360200078, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35012.1484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 499348265762816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.611328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.6531982421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8633614778518677, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772187.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2475460767745972, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8331263065338135, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.611328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.6546630859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8633631467819214, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772187.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2719526360200078, + "step": 248, + "teaching_rms": 0.27202173926514006, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3091084957122803, + "innovation_rms": 0.288511336796889, + "instruction_rms": 0.28842972417326124, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35547.5546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 449413432475648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.1689453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.4669189453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8632906675338745, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771309.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1227914094924927, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6498136520385742, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.1689453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.4683837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8632922172546387, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771309.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28842972417326124, + "step": 249, + "teaching_rms": 0.288511336796889, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3153395652770996, + "innovation_rms": 0.2949414925454183, + "instruction_rms": 0.2948874978839673, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34791.55859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 404472102649856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.6826171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.2806396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863208293914795, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770431.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0105122327804565, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4848322868347168, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.6826171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.2821044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863209843635559, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770431.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2948874978839673, + "step": 250, + "teaching_rms": 0.2949414925454183, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.290590763092041, + "innovation_rms": 0.2716782705503778, + "instruction_rms": 0.2715491157111968, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34940.1796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 364024885673984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.1572265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.0943603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8631154298782349, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769553.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9094609618186951, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3363490104675293, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.1572265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.0958251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863116979598999, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769553.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2715491157111968, + "step": 251, + "teaching_rms": 0.2716782705503778, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307459592819214, + "innovation_rms": 0.2707191889480825, + "instruction_rms": 0.2706055507265049, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35401.9921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 327622387040256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.5966796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.9080810546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8630132675170898, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768676.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8185148239135742, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2027140855789185, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.5966796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.9095458984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863014817237854, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768676.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2706055507265049, + "step": 252, + "teaching_rms": 0.2707191889480825, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3141236305236816, + "innovation_rms": 0.2754055739254139, + "instruction_rms": 0.2752728394127358, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34385.63671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 294860141625344.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.00390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.7218017578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8629027605056763, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767799.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7366633415222168, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0824426412582397, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.00390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.7232666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8629043102264404, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767799.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2752728394127358, + "step": 253, + "teaching_rms": 0.2754055739254139, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.294116258621216, + "innovation_rms": 0.28171741524698773, + "instruction_rms": 0.2816322051657996, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34097.47265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 265374134173696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773.3837890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.5355224609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8627846240997314, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766922.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6629970073699951, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9741983413696289, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773.3837890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.5369873046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8627861738204956, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766922.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2816322051657996, + "step": 254, + "teaching_rms": 0.28171741524698773, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2990899085998535, + "innovation_rms": 0.2859890019766697, + "instruction_rms": 0.2858917584020883, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33732.95703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 238836722434048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.7373046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.3492431640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8626596927642822, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766045.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5966972708702087, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8767784833908081, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.7373046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.3507080078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8626611232757568, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766045.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2858917584020883, + "step": 255, + "teaching_rms": 0.2859890019766697, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31665301322937, + "innovation_rms": 0.24368815934624374, + "instruction_rms": 0.24349149437001938, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33315.2265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 214953046835200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.068359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.1629638671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8625285625457764, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5370275378227234, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7891005873680115, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.068359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.1644287109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862529993057251, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24349149437001938, + "step": 256, + "teaching_rms": 0.24368815934624374, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3101770877838135, + "innovation_rms": 0.26744240905182903, + "instruction_rms": 0.26730449878186263, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33096.58984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 193457742151680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.3779296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.9766845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623919486999512, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764291.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4833247661590576, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7101905345916748, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.3779296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.9781494140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623933792114258, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764291.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26730449878186263, + "step": 257, + "teaching_rms": 0.26744240905182903, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3142447471618652, + "innovation_rms": 0.27951071880576106, + "instruction_rms": 0.27938617729577986, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33474.26953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 174111967936512.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770.6689453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.7904052734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8622503280639648, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763414.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4349922835826874, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6391714811325073, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770.6689453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.7918701171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862251877784729, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763414.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27938617729577986, + "step": 258, + "teaching_rms": 0.27951071880576106, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304365634918213, + "innovation_rms": 0.2811811668360158, + "instruction_rms": 0.2810756124280671, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32797.8125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 156700774498304.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.943359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.604248046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8621042966842651, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762537.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3914930522441864, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5752543210983276, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.943359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.605712890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8621058464050293, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762537.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2810756124280671, + "step": 259, + "teaching_rms": 0.2811811668360158, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3088204860687256, + "innovation_rms": 0.2649943542558928, + "instruction_rms": 0.26483423916319115, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33617.73046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141030703759360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.2021484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.4180908203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8619542121887207, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761660.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35234373807907104, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.517728865146637, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.2021484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.4195556640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8619557619094849, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761660.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26483423916319115, + "step": 260, + "teaching_rms": 0.2649943542558928, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3028502464294434, + "innovation_rms": 0.2709728894991329, + "instruction_rms": 0.270831275005025, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34452.7890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 126927633383424.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768.447265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.23193359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8618005514144897, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760783.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3171093463897705, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4659559726715088, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768.447265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.2333984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861802101135254, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760783.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.270831275005025, + "step": 261, + "teaching_rms": 0.2709728894991329, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3079583644866943, + "innovation_rms": 0.29599078257607714, + "instruction_rms": 0.2959180724950173, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35037.609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 114234872561664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.6806640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.0457763671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861643671989441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759906.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2853983938694, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41936036944389343, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.6806640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.0472412109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861645221710205, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759906.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2959180724950173, + "step": 262, + "teaching_rms": 0.29599078257607714, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3060388565063477, + "innovation_rms": 0.30885618376348234, + "instruction_rms": 0.30879741324123106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34748.1875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 102811383627776.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.9033203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.859619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8614838123321533, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759029.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25685855746269226, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37742432951927185, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.9033203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.861083984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8614853620529175, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759029.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30879741324123106, + "step": 263, + "teaching_rms": 0.30885618376348234, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.320901870727539, + "innovation_rms": 0.28049717103404825, + "instruction_rms": 0.2803658118063922, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34289.52734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 92530246942720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.1162109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.6734619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8613213300704956, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23117269575595856, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3396818935871124, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.1162109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.6749267578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8613228797912598, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758152.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2803658118063922, + "step": 264, + "teaching_rms": 0.28049717103404825, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3094470500946045, + "innovation_rms": 0.28631695112844857, + "instruction_rms": 0.2862184155626006, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34523.41796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83277218054144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.3193359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.4873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8611564636230469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757275.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20805542171001434, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3057136833667755, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.3193359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.48876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861158013343811, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757275.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2862184155626006, + "step": 265, + "teaching_rms": 0.28631695112844857, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.300950527191162, + "innovation_rms": 0.2750811971027186, + "instruction_rms": 0.2749649319510312, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34988.1015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74949494571008.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764.5146484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.3011474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8609894514083862, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756399.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18724986910820007, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2751423120498657, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764.5146484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.3026123046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8609910011291504, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756399.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2749649319510312, + "step": 266, + "teaching_rms": 0.2750811971027186, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3068981170654297, + "innovation_rms": 0.28506620633748586, + "instruction_rms": 0.2849602085039063, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34783.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67454545952768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763.703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.1151123046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8608206510543823, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755523.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1685248762369156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2476280778646469, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763.703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.1165771484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860822081565857, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755523.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2849602085039063, + "step": 267, + "teaching_rms": 0.28506620633748586, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2924373149871826, + "innovation_rms": 0.29143503177858515, + "instruction_rms": 0.2913721248033816, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33691.6484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 60709090099200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.884765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.9290771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8606500625610352, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754647.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1516723781824112, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2228652685880661, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.884765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.9305419921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8606514930725098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754647.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2913721248033816, + "step": 268, + "teaching_rms": 0.29143503177858515, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.306351900100708, + "innovation_rms": 0.2937408764765259, + "instruction_rms": 0.29365765498989604, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33461.734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 54638178992128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.060546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.7430419921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8604779243469238, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753771.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1365051418542862, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2005787342786789, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.060546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.7445068359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8604793548583984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753771.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29365765498989604, + "step": 269, + "teaching_rms": 0.2937408764765259, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.29740834236145, + "innovation_rms": 0.28253340441579106, + "instruction_rms": 0.28245497313527296, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32833.671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 49174359834624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761.2314453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.5570068359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8603044748306274, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752895.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12285462766885757, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18052086234092712, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761.2314453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.5584716796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8603057861328125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752895.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28245497313527296, + "step": 270, + "teaching_rms": 0.28253340441579106, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307465076446533, + "innovation_rms": 0.27256050082409095, + "instruction_rms": 0.2724343230838479, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33170.03515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 44256924270592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760.3974609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.3709716796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860129714012146, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752019.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11056916415691376, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1624687761068344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760.3974609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.3724365234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860131025314331, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752019.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2724343230838479, + "step": 271, + "teaching_rms": 0.27256050082409095, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3073205947875977, + "innovation_rms": 0.2716928646440787, + "instruction_rms": 0.27157304000230575, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33558.015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 39831233101824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.5595703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.1849365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8599538803100586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751143.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0995122417807579, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14622189104557037, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.5595703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.1864013671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8599551916122437, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751143.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27157304000230575, + "step": 272, + "teaching_rms": 0.2716928646440787, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307616949081421, + "innovation_rms": 0.25873359532577506, + "instruction_rms": 0.25857651879963345, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33064.77734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 35848108113920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758.7177734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.9989013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8597769737243652, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750267.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08956101536750793, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13159969449043274, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758.7177734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.0003662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8597782850265503, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750267.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25857651879963345, + "step": 273, + "teaching_rms": 0.25873359532577506, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2996084690093994, + "innovation_rms": 0.2654440730962247, + "instruction_rms": 0.2653359208440962, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32840.265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 32263297302528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.8720703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.81298828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8595991134643555, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749391.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0806049108505249, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11843971908092499, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.8720703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.814453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85960054397583, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749391.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2653359208440962, + "step": 274, + "teaching_rms": 0.2654440730962247, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2980923652648926, + "innovation_rms": 0.2820107644277317, + "instruction_rms": 0.28191254387359455, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32745.013671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 29036967362560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.0234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.6270751953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8594205379486084, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748515.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07254441827535629, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10659574717283249, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.0234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.6285400390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859421968460083, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748515.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28191254387359455, + "step": 275, + "teaching_rms": 0.2820107644277317, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3035073280334473, + "innovation_rms": 0.27097661932059575, + "instruction_rms": 0.2708429964227785, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32360.703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 26133269577728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.441162109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859241247177124, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747639.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06528997421264648, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09593617171049118, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.442626953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859242558479309, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747639.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2708429964227785, + "step": 276, + "teaching_rms": 0.27097661932059575, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304506540298462, + "innovation_rms": 0.27539353855180604, + "instruction_rms": 0.27526654116134, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32523.55078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 23519943458816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755.318359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.2552490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8590612411499023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746763.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05876097455620766, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08634255081415176, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755.318359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.2567138671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8590625524520874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746763.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27526654116134, + "step": 277, + "teaching_rms": 0.27539353855180604, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3166027069091797, + "innovation_rms": 0.25346183147305495, + "instruction_rms": 0.25329076363108294, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32863.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 21167949742080.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754.462890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.0693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.858880639076233, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05288487672805786, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07770829647779465, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754.462890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.07080078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.858881950378418, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25329076363108294, + "step": 278, + "teaching_rms": 0.25346183147305495, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.309544086456299, + "innovation_rms": 0.2959230083189856, + "instruction_rms": 0.29582757603559384, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32780.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19051153719296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753.60546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.8834228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8586995601654053, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745013.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047596387565135956, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06993746757507324, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753.60546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.8848876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8587007522583008, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745013.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29582757603559384, + "step": 279, + "teaching_rms": 0.2959230083189856, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.309030294418335, + "innovation_rms": 0.28179095130804693, + "instruction_rms": 0.28168682139794954, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32445.27734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17146037927936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.74609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.6976318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8585178852081299, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744138.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0428367480635643, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0629437193274498, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.74609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.6990966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8585190773010254, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744138.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28168682139794954, + "step": 280, + "teaching_rms": 0.28179095130804693, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3103232383728027, + "innovation_rms": 0.2785562115471031, + "instruction_rms": 0.27843269505945967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33404.78125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15431433715712.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.884765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.5118408203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8583358526229858, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743263.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03855307400226593, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0566493459045887, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.884765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.5133056640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8583370447158813, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743263.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27843269505945967, + "step": 281, + "teaching_rms": 0.2785562115471031, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308316230773926, + "innovation_rms": 0.25801942198563865, + "instruction_rms": 0.2578818488050589, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33439.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13888290553856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.0224609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.3260498046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8581534624099731, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742388.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0346977673470974, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05098440870642662, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.0224609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.3275146484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8581546545028687, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742388.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2578818488050589, + "step": 282, + "teaching_rms": 0.25801942198563865, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308720588684082, + "innovation_rms": 0.268472879389723, + "instruction_rms": 0.2683484300890217, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32784.33984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12499461079040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750.158203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.1402587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8579707145690918, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741513.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.031227989122271538, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04588596522808075, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750.158203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.1417236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8579719066619873, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741513.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2683484300890217, + "step": 283, + "teaching_rms": 0.268472879389723, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2981810569763184, + "innovation_rms": 0.2986883620037293, + "instruction_rms": 0.29862880536506686, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33639.69140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11249514446848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749.29296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.9544677734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8577876091003418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740638.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.02810519002377987, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.041297368705272675, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749.29296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.9559326171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8577888011932373, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740638.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29862880536506686, + "step": 284, + "teaching_rms": 0.2986883620037293, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3011858463287354, + "innovation_rms": 0.261882739170452, + "instruction_rms": 0.26173333505517854, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32432.119140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10124563316736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748.4267578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.7686767578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8576041460037231, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739763.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.025294670835137367, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03716763108968735, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748.4267578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.7701416015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8576054573059082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739763.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26173333505517854, + "step": 285, + "teaching_rms": 0.261882739170452, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3011229038238525, + "innovation_rms": 0.2628743676613335, + "instruction_rms": 0.26274281835685265, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31947.896484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9112106565632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.560546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.5828857421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8574204444885254, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.022765202447772026, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03345086798071861, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.560546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.5843505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.857421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26274281835685265, + "step": 286, + "teaching_rms": 0.2628743676613335, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2982208728790283, + "innovation_rms": 0.2729924114839236, + "instruction_rms": 0.2728766239255228, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32557.486328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8200896118784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746.693359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.397216796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.857236623764038, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738013.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.02048868127167225, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03010578081011772, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746.693359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.398681640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8572380542755127, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738013.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2728766239255228, + "step": 287, + "teaching_rms": 0.2729924114839236, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.30659818649292, + "innovation_rms": 0.25854979436709347, + "instruction_rms": 0.25840654650975947, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33078.0390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7380806402048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745.8251953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.2115478515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8570526838302612, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737138.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01843981258571148, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.02709520235657692, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745.8251953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.2130126953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8570539951324463, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737138.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25840654650975947, + "step": 288, + "teaching_rms": 0.25854979436709347, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3070290088653564, + "innovation_rms": 0.25505321895885164, + "instruction_rms": 0.2549115440726197, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33573.76953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6642725814272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.9560546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.02587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568685054779053, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736264.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.016595831140875816, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.024385681375861168, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.9560546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.02734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568698167800903, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736264.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2549115440726197, + "step": 289, + "teaching_rms": 0.25505321895885164, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3022401332855225, + "innovation_rms": 0.2932418180012371, + "instruction_rms": 0.2931564746868434, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34449.5703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5978453442560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.0869140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.8402099609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8566842079162598, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735390.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01493624784052372, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.021947111934423447, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.0869140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.8416748046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8566855192184448, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735390.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2931564746868434, + "step": 290, + "teaching_rms": 0.2932418180012371, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301656723022461, + "innovation_rms": 0.2755677790222609, + "instruction_rms": 0.27547077020824756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34541.47265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5380607836160.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743.216796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.654541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8564997911453247, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734516.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.013442622497677803, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.019752399995923042, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743.216796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.656005859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8565011024475098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734516.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27547077020824756, + "step": 291, + "teaching_rms": 0.2755677790222609, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.309624671936035, + "innovation_rms": 0.2813513348871347, + "instruction_rms": 0.281244932119809, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34568.4765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4842546790400.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.3466796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.4688720703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8563152551651, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733642.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.012098359875380993, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.017777159810066223, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.3466796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.4703369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8563165664672852, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733642.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.281244932119809, + "step": 292, + "teaching_rms": 0.2813513348871347, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3066153526306152, + "innovation_rms": 0.24845564314083096, + "instruction_rms": 0.24830350362944323, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34921.21484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4358292111360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741.4755859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.283203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.856130599975586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.010888523422181606, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0159994438290596, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741.4755859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.28466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8561317920684814, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24830350362944323, + "step": 293, + "teaching_rms": 0.24845564314083096, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.310338258743286, + "innovation_rms": 0.27764852112734495, + "instruction_rms": 0.2775432385483914, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34968.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3922462769152.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740.6044921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.09765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8559458255767822, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731894.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.009799670428037643, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.014399498701095581, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740.6044921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.09912109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8559470176696777, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731894.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2775432385483914, + "step": 294, + "teaching_rms": 0.27764852112734495, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3000028133392334, + "innovation_rms": 0.25550736144475517, + "instruction_rms": 0.25535681511241937, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35077.203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3530216439808.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739.7333984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.912109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855760931968689, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731020.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00881970301270485, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.012959548272192478, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739.7333984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.91357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8557621240615845, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731020.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25535681511241937, + "step": 295, + "teaching_rms": 0.25550736144475517, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301285743713379, + "innovation_rms": 0.2879855436216822, + "instruction_rms": 0.287916356482223, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36055.50390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3177194717184.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738.861328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.7265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8555760383605957, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730146.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00793773215264082, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011663593351840973, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738.861328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.72802734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8555772304534912, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730146.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.287916356482223, + "step": 296, + "teaching_rms": 0.2879855436216822, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3079209327697754, + "innovation_rms": 0.2727608084014196, + "instruction_rms": 0.2726629434830635, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36218.0546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2859475140608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.9892578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855391025543213, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.007143958937376738, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01049723383039236, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.9892578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.54248046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855392336845398, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2726629434830635, + "step": 297, + "teaching_rms": 0.2727608084014196, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3060178756713867, + "innovation_rms": 0.2630135897804151, + "instruction_rms": 0.2628882910721396, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36449.84375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2573527678976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.1171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.35546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85520601272583, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728398.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00642956281080842, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.009447510354220867, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.1171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.35693359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8552073240280151, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728398.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2628882910721396, + "step": 298, + "teaching_rms": 0.2630135897804151, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2992184162139893, + "innovation_rms": 0.2552271912255807, + "instruction_rms": 0.25508231115077906, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35710.46484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2316174884864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.2451171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.169921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8550208806991577, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727524.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005786606576293707, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.008502759039402008, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.2451171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.17138671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8550221920013428, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727524.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25508231115077906, + "step": 299, + "teaching_rms": 0.2552271912255807, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3002405166625977, + "innovation_rms": 0.28394911046120735, + "instruction_rms": 0.28384731524364426, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35427.8828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2084557422592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735.373046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.9844970703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8548357486724854, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726650.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005207945592701435, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.007652482949197292, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735.373046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.9859619140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8548370599746704, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726650.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28384731524364426, + "step": 300, + "teaching_rms": 0.28394911046120735, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.302999258041382, + "innovation_rms": 0.2768150154591984, + "instruction_rms": 0.2767285861456562, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35124.96875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1876101693440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.799072265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.854650616645813, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725777.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004687150940299034, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.006887234281748533, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.800537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.854651927947998, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725777.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2767285861456562, + "step": 301, + "teaching_rms": 0.2768150154591984, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3106067180633545, + "innovation_rms": 0.2900621462457677, + "instruction_rms": 0.2900022836049087, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34610.984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1688491524096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733.6279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.6136474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8544654846191406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004218435846269131, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.006198510527610779, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733.6279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.6151123046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8544667959213257, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724904.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2900022836049087, + "step": 302, + "teaching_rms": 0.2900621462457677, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31005859375, + "innovation_rms": 0.26684435432162296, + "instruction_rms": 0.26676355438347044, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34268.37890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1519642345472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732.755859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.42822265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8542802333831787, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724031.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0037965921219438314, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005578659474849701, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732.755859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.4296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8542815446853638, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724031.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26676355438347044, + "step": 303, + "teaching_rms": 0.26684435432162296, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2979652881622314, + "innovation_rms": 0.2497740389351926, + "instruction_rms": 0.2496351814413671, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34192.37109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1367678124032.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.8828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.2427978515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8540949821472168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723158.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003416932886466384, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005020793527364731, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.8828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.2442626953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8540964126586914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723158.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2496351814413671, + "step": 304, + "teaching_rms": 0.2497740389351926, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301604747772217, + "innovation_rms": 0.2693353395498597, + "instruction_rms": 0.2692259782619113, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33546.4453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230910259200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.009765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.057373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8539097309112549, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722285.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0030752394814044237, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004518714267760515, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.009765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.058837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8539111614227295, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722285.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2692259782619113, + "step": 305, + "teaching_rms": 0.2693353395498597, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3041038513183594, + "innovation_rms": 0.29300589607331934, + "instruction_rms": 0.2929474488406543, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33383.55859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1107819233280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730.13671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.8719482421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8537243604660034, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721412.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002767715370282531, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00406684260815382, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730.13671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.8734130859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8537259101867676, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721412.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2929474488406543, + "step": 306, + "teaching_rms": 0.29300589607331934, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.299781560897827, + "innovation_rms": 0.23126624276634947, + "instruction_rms": 0.23108656068707362, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33092.43359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 997037309952.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.2646484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.6866455078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8535391092300415, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720539.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0024909437634050846, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0036601582542061806, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.2646484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.6881103515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8535406589508057, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720539.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.23108656068707362, + "step": 307, + "teaching_rms": 0.23126624276634947, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3011538982391357, + "innovation_rms": 0.2572422240556823, + "instruction_rms": 0.25710206788495377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31994.44140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 897333592064.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728.3916015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.5013427734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85335373878479, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719666.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002241849433630705, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0032941424287855625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728.3916015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.5028076171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8533554077148438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719666.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25710206788495377, + "step": 308, + "teaching_rms": 0.2572422240556823, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3031558990478516, + "innovation_rms": 0.28833779074579907, + "instruction_rms": 0.288274531523813, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32023.662109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 807600259072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727.5185546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.3160400390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8531684875488281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718793.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0020176644902676344, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00296472804620862, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727.5185546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.3175048828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8531700372695923, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718793.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.288274531523813, + "step": 309, + "teaching_rms": 0.28833779074579907, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307535409927368, + "innovation_rms": 0.2847519063652429, + "instruction_rms": 0.28467618008819495, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33884.44140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 726840246272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726.6455078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.1307373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8529831171035767, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0018158979946747422, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002668255241587758, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726.6455078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.1322021484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8529847860336304, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28467618008819495, + "step": 310, + "teaching_rms": 0.2847519063652429, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3039276599884033, + "innovation_rms": 0.2734248341056874, + "instruction_rms": 0.27333294308844575, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33795.5703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 654156234752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725.7734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.9454345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8527978658676147, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717047.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.001634308136999607, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002401429694145918, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725.7734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.9468994140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.852799415588379, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717047.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27333294308844575, + "step": 311, + "teaching_rms": 0.2734248341056874, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3061821460723877, + "innovation_rms": 0.2699133393804393, + "instruction_rms": 0.2698119828201293, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33904.6875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 588740624384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.900390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.7601318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8526124954223633, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716175.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0014708773232996464, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002161286771297455, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.900390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.7615966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.852614164352417, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716175.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2698119828201293, + "step": 312, + "teaching_rms": 0.2699133393804393, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3083369731903076, + "innovation_rms": 0.270467607500423, + "instruction_rms": 0.27037651337540103, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34267.34375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 529866555392.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.02734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.5748291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8524272441864014, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715303.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0013237895909696817, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0019451580010354519, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.02734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.5762939453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8524287939071655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715303.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27037651337540103, + "step": 313, + "teaching_rms": 0.270467607500423, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.322415590286255, + "innovation_rms": 0.3338051136361951, + "instruction_rms": 0.33384327996122615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34830.48046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 476879912960.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723.154296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.3896484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8522419929504395, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714431.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0011914105853065848, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0017506422009319067, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723.154296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.39111328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8522435426712036, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714431.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.33384327996122615, + "step": 314, + "teaching_rms": 0.3338051136361951, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2953264713287354, + "innovation_rms": 0.2755120156247644, + "instruction_rms": 0.27542138552778467, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34438.2109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 429191921664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.2822265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.2044677734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.852056622505188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713559.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0010722695151343942, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0015755778877064586, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.2822265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.2059326171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8520581722259521, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713559.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27542138552778467, + "step": 315, + "teaching_rms": 0.2755120156247644, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3050618171691895, + "innovation_rms": 0.3016880662983754, + "instruction_rms": 0.30163777944227105, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35195.21875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 386272722944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721.4091796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.019287109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.851871371269226, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712687.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0009650425636209548, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.001418020110577345, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721.4091796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.020751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8518729209899902, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712687.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30163777944227105, + "step": 316, + "teaching_rms": 0.3016880662983754, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301314353942871, + "innovation_rms": 0.26298608826522385, + "instruction_rms": 0.26286772684863585, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35485.23046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 347645444096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720.537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.8341064453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8516861200332642, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711815.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0008685382781550288, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0012762180995196104, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720.537109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.8355712890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8516876697540283, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711815.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26286772684863585, + "step": 317, + "teaching_rms": 0.26298608826522385, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.306528091430664, + "innovation_rms": 0.2684217920832563, + "instruction_rms": 0.26832261503382776, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35866.52734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 312880889856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719.6650390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.64892578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8515008687973022, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710943.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007816844154149294, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0011485962895676494, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719.6650390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.650390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8515024185180664, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710943.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26832261503382776, + "step": 318, + "teaching_rms": 0.2684217920832563, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3064236640930176, + "innovation_rms": 0.28972672791674414, + "instruction_rms": 0.2896613653769902, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36255.71484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 281592791040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718.79296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.4637451171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8513156175613403, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710071.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007035159505903721, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0010337366256862879, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718.79296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.4652099609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8513171672821045, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710071.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2896613653769902, + "step": 319, + "teaching_rms": 0.28972672791674414, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2994956970214844, + "innovation_rms": 0.30577656229897526, + "instruction_rms": 0.3057195223932587, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36013.109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 253433511936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.9208984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.2786865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.851130485534668, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709199.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0006331643671728671, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0009303629631176591, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.9208984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.2801513671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8511319160461426, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709199.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3057195223932587, + "step": 320, + "teaching_rms": 0.30577656229897526, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304568290710449, + "innovation_rms": 0.26431433615154226, + "instruction_rms": 0.26420978220102487, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35666.6328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 228090167296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.0478515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.0936279296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.850945234298706, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708327.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0005698479362763464, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0008373266318812966, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.0478515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.0950927734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8509467840194702, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708327.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26420978220102487, + "step": 321, + "teaching_rms": 0.26431433615154226, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.298562526702881, + "innovation_rms": 0.2840294319406367, + "instruction_rms": 0.28394392998665857, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35487.015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 205281148928.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716.17578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.9085693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8507599830627441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707455.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0005128631019033492, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007535939221270382, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716.17578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.9100341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8507615327835083, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707455.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28394392998665857, + "step": 322, + "teaching_rms": 0.2840294319406367, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304377794265747, + "innovation_rms": 0.25619865800130814, + "instruction_rms": 0.2560767242247034, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36119.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 184753029120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715.3037109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.7235107421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8505748510360718, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706583.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0004615767684299499, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000678234500810504, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715.3037109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.7249755859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8505762815475464, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706583.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2560767242247034, + "step": 323, + "teaching_rms": 0.25619865800130814, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308640718460083, + "innovation_rms": 0.2829524003225242, + "instruction_rms": 0.2828529033989976, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36518.265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 166277726208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.431640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.5384521484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8503897190093994, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00041541908285580575, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0006104110507294536, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.431640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.5399169921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.850391149520874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705712.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2828529033989976, + "step": 324, + "teaching_rms": 0.2829524003225242, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.29638409614563, + "innovation_rms": 0.2886440562412176, + "instruction_rms": 0.2885618704625799, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36136.7109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 149649948672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713.5595703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.3533935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8502044677734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704841.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003738771774806082, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0005493699572980404, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713.5595703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.3548583984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8502060174942017, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704841.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2885618704625799, + "step": 325, + "teaching_rms": 0.2886440562412176, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308162212371826, + "innovation_rms": 0.28761125887870703, + "instruction_rms": 0.2875001420595795, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35795.26171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 134684950528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.1683349609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8500193357467651, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703970.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00033648943644948304, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0004944329266436398, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.1697998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8500207662582397, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703970.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2875001420595795, + "step": 326, + "teaching_rms": 0.28761125887870703, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.319580554962158, + "innovation_rms": 0.2784556520784294, + "instruction_rms": 0.278348337911314, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36360.140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 121216458752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711.8154296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.9833984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8498342037200928, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703099.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003028404898941517, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00044498962233774364, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711.8154296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.98486328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8498356342315674, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703099.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.278348337911314, + "step": 327, + "teaching_rms": 0.2784556520784294, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.300400495529175, + "innovation_rms": 0.27360456222717955, + "instruction_rms": 0.27348621005629936, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35162.87890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 109094813696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.943359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.7984619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8496490716934204, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702228.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0002725564409047365, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00040049065137282014, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.943359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.7999267578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.849650502204895, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702228.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27348621005629936, + "step": 328, + "teaching_rms": 0.27360456222717955, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.297814130783081, + "innovation_rms": 0.2632616238292976, + "instruction_rms": 0.2631475324358249, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34813.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 98185330688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.0712890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.613525390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8494640588760376, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701357.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0002453007909934968, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00036044156877323985, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.0712890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.614990234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8494653701782227, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701357.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2631475324358249, + "step": 329, + "teaching_rms": 0.2632616238292976, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3162481784820557, + "innovation_rms": 0.25657891046474496, + "instruction_rms": 0.2564170474954674, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34452.4921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 88366800896.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709.19921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.4285888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8492789268493652, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700486.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00022077070025261492, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003243974060751498, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709.19921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.4300537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8492803573608398, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700486.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2564170474954674, + "step": 330, + "teaching_rms": 0.25657891046474496, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.303004741668701, + "innovation_rms": 0.26801417412158485, + "instruction_rms": 0.2678893762942696, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34831.4609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79530123264.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708.328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.24365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8490937948226929, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699615.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00019869362586177886, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0002919576654676348, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708.328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.2451171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.849095344543457, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699615.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2678893762942696, + "step": 331, + "teaching_rms": 0.26801417412158485, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3040096759796143, + "innovation_rms": 0.28959446885988255, + "instruction_rms": 0.28952445783121783, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35646.390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71577108480.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707.45703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.0587158203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8489086627960205, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0001788242516340688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00026276189601048827, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707.45703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.0601806640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8489103317260742, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28952445783121783, + "step": 332, + "teaching_rms": 0.28959446885988255, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3032758235931396, + "innovation_rms": 0.2788685639031087, + "instruction_rms": 0.2787727130123362, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36826.171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64419397632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706.5849609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.8739013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8487236499786377, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697873.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00016094182501547039, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00023648570640943944, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706.5849609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.8753662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8487251996994019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697873.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2787727130123362, + "step": 333, + "teaching_rms": 0.2788685639031087, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.305079698562622, + "innovation_rms": 0.27920223141420836, + "instruction_rms": 0.2790831980520559, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36708.72265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 57977458688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705.7138671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.6890869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8485386371612549, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697002.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0001448476396035403, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00021283712703734636, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705.7138671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.6905517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.848540186882019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697002.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2790831980520559, + "step": 334, + "teaching_rms": 0.27920223141420836, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3044536113739014, + "innovation_rms": 0.2826387571784897, + "instruction_rms": 0.2825424242620217, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36661.52734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 52179714048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.841796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.5042724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.848353624343872, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696132.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00013036286691203713, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00019155340851284564, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.841796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.5057373046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8483551740646362, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696132.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2825424242620217, + "step": 335, + "teaching_rms": 0.2826387571784897, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.319319486618042, + "innovation_rms": 0.2617725870451568, + "instruction_rms": 0.261639136719563, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36241.953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46961741824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.970703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.3194580078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8481686115264893, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695262.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00011732657731045038, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00017239806766156107, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.970703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.3209228515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8481701612472534, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695262.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.261639136719563, + "step": 336, + "teaching_rms": 0.2617725870451568, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.302999496459961, + "innovation_rms": 0.2714271961521836, + "instruction_rms": 0.27130711089729176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37043.05859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 42265567232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.099609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.1346435546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8479835987091064, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00010559391375863925, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00015515825361944735, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.099609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.1361083984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8479851484298706, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27130711089729176, + "step": 337, + "teaching_rms": 0.2714271961521836, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31793212890625, + "innovation_rms": 0.2484898318317899, + "instruction_rms": 0.2483288116997012, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35756.82421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38039011328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702.228515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.9498291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8477987051010132, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693522.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.503452019998804e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00013964241952635348, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702.228515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.9512939453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8478001356124878, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693522.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2483288116997012, + "step": 338, + "teaching_rms": 0.2484898318317899, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3050150871276855, + "innovation_rms": 0.24607951874007386, + "instruction_rms": 0.24589755054209447, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35738.0078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34235109376.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701.357421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.7650146484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.84761381149292, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692652.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.553106454201043e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000125678168842569, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701.357421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.7664794921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8476152420043945, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692652.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24589755054209447, + "step": 339, + "teaching_rms": 0.24607951874007386, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3007845878601074, + "innovation_rms": 0.27806233327575625, + "instruction_rms": 0.2779822862528298, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35890.66015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 30811598848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700.4873046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.580322265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8474289178848267, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691782.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.697795808780938e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00011311034904792905, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700.4873046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.581787109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8474303483963013, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691782.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2779822862528298, + "step": 340, + "teaching_rms": 0.27806233327575625, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.314487934112549, + "innovation_rms": 0.22149206750227898, + "instruction_rms": 0.22125733477944715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35494.07421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 27730438144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699.6162109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.3956298828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8472440242767334, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.928016227902845e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00010179931268794462, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699.6162109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.3970947265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.847245454788208, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.22125733477944715, + "step": 341, + "teaching_rms": 0.22149206750227898, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.303394079208374, + "innovation_rms": 0.2772566090447132, + "instruction_rms": 0.2771386569658654, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34833.43359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24957394944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698.7451171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.2109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8470591306686401, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690042.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.235214095795527e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.161938214674592e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698.7451171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.21240234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8470605611801147, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690042.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2771386569658654, + "step": 342, + "teaching_rms": 0.2772566090447132, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3023366928100586, + "innovation_rms": 0.2804385713347286, + "instruction_rms": 0.2803486557101565, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35683.1328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22461655040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.0262451171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8468742370605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8689172.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.611692540696822e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.245743811130524e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.0277099609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8468756675720215, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8689172.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2803486557101565, + "step": 343, + "teaching_rms": 0.2804385713347286, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.298776865005493, + "innovation_rms": 0.26250410348192543, + "instruction_rms": 0.262372941410382, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36267.2890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20215488512.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.0048828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.841552734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8466893434524536, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8688302.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0505230319686234e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.421168993460014e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.0048828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.843017578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8466907739639282, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8688302.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.262372941410382, + "step": 344, + "teaching_rms": 0.26250410348192543, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.302884340286255, + "innovation_rms": 0.28336259154925825, + "instruction_rms": 0.2832701367846047, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35647.71484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19763480576.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696.134765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.6568603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8465044498443604, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8687432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5454704377334565e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.679051875835285e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696.134765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.6583251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.846505880355835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8687432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2832701367846047, + "step": 345, + "teaching_rms": 0.28336259154925825, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2959072589874268, + "innovation_rms": 0.2576541098958782, + "instruction_rms": 0.257547023226699, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35653.3515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19728865280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695.263671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.47216796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.846319556236267, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8686562.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0909231756813824e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.01114661549218e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695.263671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.4736328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8463211059570312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8686562.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.257547023226699, + "step": 346, + "teaching_rms": 0.2576541098958782, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3029963970184326, + "innovation_rms": 0.2647267797980056, + "instruction_rms": 0.2645975324370069, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36131.390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19810314240.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.3935546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.28759765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8461347818374634, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8685693.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.681830639834516e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4100317356642336e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.3935546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.2890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8461363315582275, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8685693.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2645975324370069, + "step": 347, + "teaching_rms": 0.2647267797980056, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3057830333709717, + "innovation_rms": 0.25457121986908426, + "instruction_rms": 0.25443099794438656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35668.6796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19953399808.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693.5234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.10302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8459501266479492, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8684824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.313647539471276e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.8690282710595056e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693.5234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.1044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8459515571594238, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8684824.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25443099794438656, + "step": 348, + "teaching_rms": 0.25457121986908426, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3068795204162598, + "innovation_rms": 0.2829021781292106, + "instruction_rms": 0.2828062739936588, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35843.70703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19933077504.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692.6533203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.91845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8457653522491455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683955.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9822827855241485e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3821251892950386e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692.6533203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.919921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8457667827606201, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683955.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2828062739936588, + "step": 349, + "teaching_rms": 0.2829021781292106, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.303715229034424, + "innovation_rms": 0.2747147069271108, + "instruction_rms": 0.27461643421075116, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36015.66796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19901931520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691.783203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.73388671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8455805778503418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683086.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6840543796424754e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.943912452086806e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691.783203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.7353515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8455820083618164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683086.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27461643421075116, + "step": 350, + "teaching_rms": 0.2747147069271108, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308358907699585, + "innovation_rms": 0.29625551387412313, + "instruction_rms": 0.2961951705813059, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35441.57421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19939577856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.9130859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.54931640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.845395803451538, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682217.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4156488507287577e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.549520988599397e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.9130859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.55078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8453972339630127, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682217.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2961951705813059, + "step": 351, + "teaching_rms": 0.29625551387412313, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2961888313293457, + "innovation_rms": 0.2515914459119246, + "instruction_rms": 0.2514413205977138, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34457.14453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19918106624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.04296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.36474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8452110290527344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681348.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.174083965655882e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1945688533596694e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.04296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.3662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.845212459564209, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681348.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2514413205977138, + "step": 352, + "teaching_rms": 0.2515914459119246, + "traffic_rms": 0.0 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} diff --git a/results/kp_innovation_stability_grid/margin_0p03.json b/results/kp_innovation_stability_grid/margin_0p03.json new file mode 100644 index 0000000..4fd80d2 --- /dev/null +++ b/results/kp_innovation_stability_grid/margin_0p03.json @@ -0,0 +1,32986 @@ +{ + "first_any_nonfinite_groups": {}, + "first_any_nonfinite_step": null, + "first_nonfinite_by_group": {}, + "first_nonfinite_by_metric": {}, + "first_nonfinite_metrics": [], + "first_training_failure_groups": {}, + "first_training_failure_step": null, + "initial_parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4011031687259674, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3956821858882904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1162850856781006, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 0.03337550908327103, + "max_applied_stability_margin": 0.033375516533851624, + "max_positive_residual_soma_slope": 0.0, + "min_residual_soma_slope": -0.03337550908327103, + "mse": 0.0007272035929971659, + "observations": 64, + "residual_traffic_rms_ratio": 1.5995902476703123, + "stability_margin": 0.03 + }, + "first_mse": 0.0007272035929971659, + "last_mse": 0.0007272035929971659, + "post_warmup_traffic_residual_rms_ratio": 1.5995902476703123, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_mixed_traffic_nonfinite_diagnosis_v1", + "provenance": { + "git_commit": "0ae690e7b3144da2c5b3384cd7169e86529fd26f", + "git_tracked_dirty": false + }, + "rule": "innovation", + "scope": "training_only_no_validation_or_test_evaluation", + "split": { + "dataset": "cifar10", + "input_layout": "NCHW", + "input_shape": [ + 3, + 32, + 32 + ], + "loader_seed": 0, + "normalization_mean": [ + 0.49140000343322754, + 0.4821999967098236, + 0.4465000033378601 + ], + "normalization_std": [ + 0.24699999392032623, + 0.2434999942779541, + 0.26159998774528503 + ], + "split_from_training_only": true, + "split_seed": 2027, + "test_examples": 10000, + "train_examples": 45000, + "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5", + "validation_class_counts": { + "0": 500, + "1": 500, + "2": 500, + "3": 500, + "4": 500, + "5": 500, + "6": 500, + "7": 500, + "8": 500, + "9": 500 + }, + "validation_examples": 5000, + "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b" + }, + "stability_margin": 0.03, + "test_evaluations": 0, + "traffic_calibration": { + "instruction_rms": [ + 0.008471464738249779, + 0.006617916747927666, + 0.003988116513937712, + 0.0043865954503417015, + 0.002713605063036084, + 0.003158062230795622, + 0.002004207344725728, + 0.003757704282179475, + 0.0027085733599960804, + 0.0030200635083019733, + 0.002056022873148322, + 0.002389939734712243, + 0.0016842411132529378, + 0.0034010226372629404, + 0.002549725119024515, + 0.0030542907770723104, + 0.0020918569061905146, + 0.002605808200314641, + 0.0018916188273578882 + ], + "realized_traffic_instruction_rms_ratio": [ + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 3.999999523162842, + 4.0, + 4.000000476837158, + 3.999999761581421, + 4.0, + 4.0, + 3.999999761581421, + 4.0, + 4.000000476837158, + 4.000000476837158, + 4.0, + 4.0, + 4.0 + ], + "target_ratio": 4.0, + "traffic_gain": [ + 0.0443243607878685, + 0.035817068070173264, + 0.013778294436633587, + 0.022762222215533257, + 0.007214079145342112, + 0.017218582332134247, + 0.004616298712790012, + 0.02042512409389019, + 0.007327491883188486, + 0.01611751690506935, + 0.004703814629465342, + 0.01256294921040535, + 0.003324519144371152, + 0.01805141195654869, + 0.006063441745936871, + 0.016296451911330223, + 0.0044285994954407215, + 0.013874167576432228, + 0.0036007820162922144 + ], + "unscaled_traffic_rms": [ + 0.7644973993301392, + 0.7390796542167664, + 1.157796859741211, + 0.7708553671836853, + 1.504616141319275, + 0.7336404919624329, + 1.736635684967041, + 0.7358984351158142, + 1.478581428527832, + 0.7495108246803284, + 1.7483876943588257, + 0.7609485983848572, + 2.0264477729797363, + 0.7536302804946899, + 1.6820316314697266, + 0.7496824264526367, + 1.8894071578979492, + 0.7512690424919128, + 2.10134220123291 + ] + }, + "trajectory": [ + { + "batch_loss": 2.9134020805358887, + "innovation_rms": 0.027791247511226837, + "instruction_rms": 0.004215397890235521, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40285658836364746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9875131845474243, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.124950408935547, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4058851897716522, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6322960257530212, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7759182453155518, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1675523519515991, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.67552375793457, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.870853066444397, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.70853042602539, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.124971389770508, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3794170618057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.632293164730072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.757441282272339, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017318262576895525, + "step": 1, + "teaching_rms": 0.027791247511226837, + "traffic_rms": 0.01680093259933294 + }, + { + "batch_loss": 2.306366443634033, + "innovation_rms": 0.2742800857960511, + "instruction_rms": 0.2727156771548255, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36302483081817627, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 159.16864013671875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.28965950012207, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4102132022380829, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5693559646606445, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.304884433746338, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.376168966293335, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.399652481079102, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.727872848510742, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.75712585449219, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.28969955444336, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.387024462223053, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.56935054063797, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.233582496643066, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2727708320848526, + "step": 2, + "teaching_rms": 0.2742800857960511, + "traffic_rms": 0.005487565935528707 + }, + { + "batch_loss": 2.491521120071411, + "innovation_rms": 0.41671248531981364, + "instruction_rms": 0.41603139938179945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.903782606124878, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 569.750244140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.760162353515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4162951409816742, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4571085274219513, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.5809006690979, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.486708641052246, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.103092193603516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.09678840637207, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.68915939331055, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.76021957397461, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4192211627960205, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4571007490158081, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.462057113647461, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.4161360134718316, + "step": 3, + "teaching_rms": 0.41671248531981364, + "traffic_rms": 0.009329334142785225 + }, + { + "batch_loss": 3.1429684162139893, + "innovation_rms": 0.1158643752561682, + "instruction_rms": 0.10390036522525907, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.350889205932617, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2275.425048828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.483388900756836, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42041441798210144, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.613958477973938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.629240036010742, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.5017499923706055, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.53070068359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.809185028076172, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 245.2412109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.483461380004883, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43143633008003235, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.614013671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.467609405517578, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.1046479262229814, + "step": 4, + "teaching_rms": 0.1158643752561682, + "traffic_rms": 0.012500514106978803 + }, + { + "batch_loss": 3.19339919090271, + "innovation_rms": 58.235019530769684, + "instruction_rms": 58.23502365994121, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.714153289794922, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3589.557373046875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.619752883911133, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4203285276889801, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4958186149597168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.472648620605469, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.470046997070312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.26422882080078, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.93264389038086, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.7170867919922, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.619718551635742, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4264024794101715, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49588438868522644, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.272512435913086, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 58.23502402371302, + "step": 5, + "teaching_rms": 58.235019530769684, + "traffic_rms": 0.006512936499391681 + }, + { + "batch_loss": 3.537811279296875, + "innovation_rms": 73.7926624976436, + "instruction_rms": 73.79245564113647, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.04273796081543, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 49540.1484375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.5781478881836, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41928818821907043, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5724515318870544, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.079181671142578, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.74576473236084, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.757179260253906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.73013687133789, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 936.6278076171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.5781021118164, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40073075890541077, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5723889470100403, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.866990089416504, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 73.79246161674129, + "step": 6, + "teaching_rms": 73.7926624976436, + "traffic_rms": 0.02969872165885092 + }, + { + "batch_loss": 3.240792751312256, + "innovation_rms": 0.48002596093466, + "instruction_rms": 0.4786552261801404, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.991260528564453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 44619.125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.21099090576172, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.418327271938324, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7266679406166077, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.22978973388672, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.593910217285156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.383060455322266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.02664184570312, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 842.9650268554688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.21094512939453, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40254655480384827, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7268329858779907, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.331466674804688, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.4787203509413934, + "step": 7, + "teaching_rms": 0.48002596093466, + "traffic_rms": 0.007898683708078637 + }, + { + "batch_loss": 3.735729694366455, + "innovation_rms": 13.730580066377033, + "instruction_rms": 13.73016479270646, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.205096244812012, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 212669.921875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.8477172851562, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41807821393013, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9059659242630005, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.23104858398438, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.373794555664062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.79884338378906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.89349365234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 758.6685180664062, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.8477172851562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39895960688591003, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9059500694274902, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.23650360107422, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13.730184271046792, + "step": 8, + "teaching_rms": 13.730580066377033, + "traffic_rms": 0.023129084732984695 + }, + { + "batch_loss": 2.9522016048431396, + "innovation_rms": 12.792351409203924, + "instruction_rms": 12.790617320539493, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.6671028137207, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1237373.125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2613.992431640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46945905685424805, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5507502555847168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 261.17010498046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.50479507446289, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.30999755859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.17364501953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 682.8016357421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2613.992431640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4438917338848114, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.550733208656311, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 261.1007385253906, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12.790702916834718, + "step": 9, + "teaching_rms": 12.792351409203924, + "traffic_rms": 0.04679590655215282 + }, + { + "batch_loss": 4.459118366241455, + "innovation_rms": 1.248058394659339, + "instruction_rms": 1.238184604239556, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.624603271484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1198244.625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4072.13720703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5493706464767456, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3090943098068237, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 496.4267883300781, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.46990966796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.65115356445312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 345.19439697265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1122.7869873046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4072.13720703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4700230360031128, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3090760707855225, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 496.357421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.238506554355396, + "step": 10, + "teaching_rms": 1.248058394659339, + "traffic_rms": 0.028272771096332937 + }, + { + "batch_loss": 13.07931900024414, + "innovation_rms": 4932.03702753456, + "instruction_rms": 4932.037027051426, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.465187072753906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1078420.125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3664.882568359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6687363982200623, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.480172157287598, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 774.8396606445312, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.03851318359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158.4014434814453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 399.21307373046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010.5082397460938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3664.882568359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7183689475059509, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.480079650878906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 774.873291015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4932.037027061946, + "step": 11, + "teaching_rms": 4932.03702753456, + "traffic_rms": 0.010286709689369223 + }, + { + "batch_loss": 8.458208084106445, + "innovation_rms": 4679.551224859599, + "instruction_rms": 4679.551372504901, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.41870880126953, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 970578.125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3298.31689453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4372737407684326, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.583579063415527, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1104.67138671875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.35025787353516, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 237.4093475341797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 447.8298645019531, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 909.4573974609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3298.31689453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5715689659118652, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.58348274230957, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1104.7049560546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4679.551372524324, + "step": 12, + "teaching_rms": 4679.551224859599, + "traffic_rms": 0.01354655256274984 + }, + { + "batch_loss": 7.552237510681152, + "innovation_rms": 484.4487617224929, + "instruction_rms": 484.44852957174726, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 126.60831451416016, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 230681280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29558.529296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2051479816436768, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.425140380859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2955.213134765625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.53083038330078, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 229.4764862060547, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 491.5849914550781, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1233.6102294921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29558.529296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.339433193206787, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.425041198730469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2955.350341796875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 484.44853856403, + "step": 13, + "teaching_rms": 484.4487617224929, + "traffic_rms": 0.09334650881356128 + }, + { + "batch_loss": 12.546594619750977, + "innovation_rms": 26657.321958938963, + "instruction_rms": 26657.321958870012, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.10694122314453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 207613152.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26602.380859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8962128162384033, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.374384880065918, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5615.451171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.69334411621094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 284.9434814453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 530.964599609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1110.2491455078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26602.380859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.030487537384033, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.374281883239746, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5615.58837890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26657.321958871395, + "step": 14, + "teaching_rms": 26657.321958938963, + "traffic_rms": 0.008714262095196016 + }, + { + "batch_loss": 7.711289405822754, + "innovation_rms": 99711.09270196679, + "instruction_rms": 99711.09270195599, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 143.34132385253906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 186851840.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23941.580078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.626417398452759, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.436667442321777, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8009.609375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.3781509399414, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 428.03607177734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 566.40625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 999.2241821289062, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23941.580078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7603673934936523, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.436561584472656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8009.74658203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 99711.09270195618, + "step": 15, + "teaching_rms": 99711.09270196679, + "traffic_rms": 0.006344974499454643 + }, + { + "batch_loss": 17.161775588989258, + "innovation_rms": 8.648362676687954, + "instruction_rms": 8.537200206475967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 347.6746826171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 450530496.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150399.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.38568115234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.815722942352295, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15042.5205078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.9202423095703, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 395.4209289550781, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 684.8760986328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11728.5146484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150399.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.519620418548584, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.815982341766357, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15042.4228515625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8.542718178030311, + "step": 16, + "teaching_rms": 8.648362676687954, + "traffic_rms": 0.30700418272848795 + }, + { + "batch_loss": 155.881103515625, + "innovation_rms": 3160.10644287658, + "instruction_rms": 3160.1065239143213, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 312.9072265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 405477440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135357.765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.600763320922852, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97.68192291259766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28578.296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 272.49755859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1245.773193359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1740.4423828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10555.6630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135357.765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.672464370727539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97.6818618774414, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28578.19921875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3160.1065245371965, + "step": 17, + "teaching_rms": 3160.10644287658, + "traffic_rms": 0.06302831096620834 + }, + { + "batch_loss": 34.335453033447266, + "innovation_rms": 3899.576346666944, + "instruction_rms": 3899.57634264128, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 374.2159729003906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 364929696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121819.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.391141891479492, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.90377807617188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40760.2109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.1305236816406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1456.3297119140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2690.4521484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9500.0966796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121819.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.46283531188965, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.90371704101562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40760.11328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3899.5763427110805, + "step": 18, + "teaching_rms": 3899.576346666944, + "traffic_rms": 0.023461197951285446 + }, + { + "batch_loss": 21.595521926879883, + "innovation_rms": 474504.6493998512, + "instruction_rms": 474504.64939984836, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 364.6383361816406, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 328436736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109633.1328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.30230712890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.11165618896484, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 51723.5234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 549.2001953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1310.6966552734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3545.4609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8550.0869140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109633.1328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.373994827270508, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.1115951538086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 51723.42578125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 474504.64939984825, + "step": 19, + "teaching_rms": 474504.6493998512, + "traffic_rms": 0.006514353146222026 + }, + { + "batch_loss": 16.156450271606445, + "innovation_rms": 2397.2265999838087, + "instruction_rms": 2397.2265271104934, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 480.5095520019531, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 295593056.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98664.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.42210388183594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.19795227050781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61589.98828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 667.1629028320312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1179.626953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4314.96875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7695.078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98664.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.4937858581543, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.19789123535156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61589.890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2397.226527119289, + "step": 20, + "teaching_rms": 2397.2265999838087, + "traffic_rms": 0.0067449517638815985 + }, + { + "batch_loss": 26.31068229675293, + "innovation_rms": 167740.33238212974, + "instruction_rms": 167740.33238212366, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 577.4423217773438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 266033744.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 88792.0234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.81120300292969, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.8909797668457, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 70469.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 773.329345703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1061.6641845703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5007.52587890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6925.5703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 88792.0234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.882877349853516, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63.89092254638672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 70469.09375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 167740.33238212374, + "step": 21, + "teaching_rms": 167740.33238212974, + "traffic_rms": 0.006438154068927215 + }, + { + "batch_loss": 11.615644454956055, + "innovation_rms": 71531.02826375452, + "instruction_rms": 71531.02577409564, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 622.8534545898438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 239430368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79905.7734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.561004638671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.49800109863281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 78459.765625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 868.879150390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 955.4977416992188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5630.8271484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6233.01318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 79905.7734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.63267135620117, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.49794006347656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 78459.671875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 71531.02577409621, + "step": 22, + "teaching_rms": 71531.02826375452, + "traffic_rms": 0.00914450637567987 + }, + { + "batch_loss": 57.112552642822266, + "innovation_rms": 891441.6235793108, + "instruction_rms": 891441.6235793101, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 845.2896118164062, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 215487328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71907.3515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.67181396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.10808181762695, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 85650.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 954.8739624023438, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 859.9479370117188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6191.79833984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5609.7119140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71907.3515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.743473052978516, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.1080207824707, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 85650.40625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 891441.6235793101, + "step": 23, + "teaching_rms": 891441.6235793108, + "traffic_rms": 0.008005296037976888 + }, + { + "batch_loss": 10.513875007629395, + "innovation_rms": 188469.72392814892, + "instruction_rms": 188469.72392814409, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1616.259033203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 193938592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64708.046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.27104568481445, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45.992305755615234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92121.3046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1032.269287109375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 773.953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6696.67236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5048.74072265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64708.046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.34269714355469, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45.992244720458984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92121.2109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 188469.7239281443, + "step": 24, + "teaching_rms": 188469.72392814892, + "traffic_rms": 0.008207758892832828 + }, + { + "batch_loss": 51.883277893066406, + "innovation_rms": 131110.51608994542, + "instruction_rms": 131110.5160899247, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2697.659912109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 174544736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58228.03125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.40927505493164, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.38229751586914, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97944.109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1101.925048828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 696.5578002929688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7151.05908203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4543.86669921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58228.03125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.48092269897461, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.38223648071289, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97944.015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 131110.5160899254, + "step": 25, + "teaching_rms": 131110.51608994542, + "traffic_rms": 0.013688997751832032 + }, + { + "batch_loss": 9.0996675491333, + "innovation_rms": 377468.86071374227, + "instruction_rms": 377468.86071374035, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3621.967529296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 157090256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52395.43359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.13230895996094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.230350494384766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103183.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1164.615234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 626.9019775390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7560.00732421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4089.47998046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52395.43359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.20395278930664, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.23028564453125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103183.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 377468.8607137406, + "step": 26, + "teaching_rms": 377468.86071374227, + "traffic_rms": 0.010494789055147346 + }, + { + "batch_loss": 4.8094682693481445, + "innovation_rms": 199763.7817586644, + "instruction_rms": 199763.78175866228, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3726.2841796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141381232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47145.5703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.482421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.5010986328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107898.2109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1221.036376953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 564.2117919921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7928.060546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3680.531982421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47145.57421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.5540542602539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.501033782958984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107898.1171875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 199763.78175866255, + "step": 27, + "teaching_rms": 199763.7817586644, + "traffic_rms": 0.00954790396445418 + }, + { + "batch_loss": 126.79713439941406, + "innovation_rms": 17174.074657231813, + "instruction_rms": 17174.07411807726, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21018.064453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12465562714112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 424094.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.49686431884766, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.981548309326172, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112140.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1291.1641845703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701.2784423828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8259.30859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19507.6953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 424094.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.56848907470703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.981475830078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112140.140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17174.074171363012, + "step": 28, + "teaching_rms": 17174.074657231813, + "traffic_rms": 1.3528792572110375 + }, + { + "batch_loss": 8293.240234375, + "innovation_rms": 522648.45853410865, + "instruction_rms": 522648.4581837878, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1841935.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 394118714884096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1128437.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 241.95191955566406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2447.048095703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115956.9296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11560.9482421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112354.6015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2763.4462890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83781.0625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1128437.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.38111877441406, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2447.0478515625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115956.8359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 522648.4582074745, + "step": 29, + "teaching_rms": 522648.45853410865, + "traffic_rms": 4.976345745394231 + }, + { + "batch_loss": 3555.84521484375, + "innovation_rms": 8.722347938624939, + "instruction_rms": 1.8094406403532988, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10405205.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 354710544449536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1015584.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 462.183837890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2202.319091796875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214405.734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21672.86328125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101119.140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7659.09326171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75402.953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1015584.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 462.6130065917969, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2202.31884765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 214405.890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.3442656479893595, + "step": 30, + "teaching_rms": 8.722347938624939, + "traffic_rms": 2.8128056056991073 + }, + { + "batch_loss": 1037.2821044921875, + "innovation_rms": 2834571.876946507, + "instruction_rms": 2834571.8769464884, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9364684.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 319239483293696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 914004.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 660.387939453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1982.0408935546875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 305806.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30773.5859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91007.2265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14445.359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67862.65625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 914004.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 660.8170776367188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1982.0406494140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 305806.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2834571.8769464884, + "step": 31, + "teaching_rms": 2834571.876946507, + "traffic_rms": 0.04670209838729726 + }, + { + "batch_loss": 243.35821533203125, + "innovation_rms": 599383.8765332766, + "instruction_rms": 599383.8765324806, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8428216.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 287315528253440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 822573.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 838.7650146484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1783.770751953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 388063.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 38964.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81906.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20552.998046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61076.390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 822573.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 839.1941528320312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1783.7705078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 388063.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 599383.8765325262, + "step": 32, + "teaching_rms": 599383.8765332766, + "traffic_rms": 0.23721410284644762 + }, + { + "batch_loss": 10615.115234375, + "innovation_rms": 108215019.80915613, + "instruction_rms": 108215019.80914406, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7585394.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0661983152399974e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 740277.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 999.2960205078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1605.309814453125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 462091.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 46335.8203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73715.8515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 26049.873046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 372321.40625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 740277.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 999.72509765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1605.3095703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 462091.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 108215019.80914456, + "step": 33, + "teaching_rms": 108215019.80915613, + "traffic_rms": 10.36226978340852 + }, + { + "batch_loss": 80380.6953125, + "innovation_rms": 125778.5302708629, + "instruction_rms": 125778.5258924297, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 69847112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.988305803458642e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 974258.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1140.424072265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1411.2799072265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 528711.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 59943.7421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160281.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 44454.01171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206770.46875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 974258.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1140.85302734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1411.2796630859375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 528711.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 125778.52607337348, + "step": 34, + "teaching_rms": 125778.5302708629, + "traffic_rms": 6.748812725873796 + }, + { + "batch_loss": 521945.8125, + "innovation_rms": 52323267396.659195, + "instruction_rms": 52323267396.659195, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 62862400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7894752746523853e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4971634.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1267.4278564453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4198.9521484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 588664.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 80431.640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 459717.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 28009.716796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 694392.375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4971634.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1267.8568115234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4198.9521484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 588664.9375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52323267396.659195, + "step": 35, + "teaching_rms": 52323267396.659195, + "traffic_rms": 19.66077170355208 + }, + { + "batch_loss": 271087.25, + "innovation_rms": 1709036255762.931, + "instruction_rms": 1709036255762.931, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56576160.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6105276956475392e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4469686.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1381.7186279296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3595.8193359375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 944118.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 121806.171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 413745.28125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 87480.5390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 624953.125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4469686.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1382.1474609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3595.8193359375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 944118.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1709036255762.931, + "step": 36, + "teaching_rms": 1709036255762.931, + "traffic_rms": 6.7912284770662845 + }, + { + "batch_loss": 9415.4609375, + "innovation_rms": 1644879652.2061746, + "instruction_rms": 1644879652.2061746, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 50918544.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4494748401834394e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4022623.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1484.5665283203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3231.38427734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1346380.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 159043.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 372370.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 143726.328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 562457.8125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4022623.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1484.9952392578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3231.38427734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1346380.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1644879652.2061746, + "step": 37, + "teaching_rms": 1644879652.2061746, + "traffic_rms": 1.187229421346629 + }, + { + "batch_loss": 637.241943359375, + "innovation_rms": 1096235934.372859, + "instruction_rms": 1096235934.372859, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 45826688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3045273389852262e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3620226.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1577.11474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2908.143798828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1708403.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 192556.609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 335133.65625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 194347.53125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 506212.03125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3620226.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1577.54345703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2908.143798828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1708403.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1096235934.372859, + "step": 38, + "teaching_rms": 1096235934.372859, + "traffic_rms": 3.9002592756564134 + }, + { + "batch_loss": 7307.40673828125, + "innovation_rms": 281973332.8471675, + "instruction_rms": 281973332.8471674, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 41244020.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1740745793168998e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3258032.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1660.392333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2617.197998046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2034206.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 222718.640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 301620.28125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 239906.609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 455590.8125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3258032.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1660.821044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2617.197998046875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2034206.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 281973332.8471674, + "step": 39, + "teaching_rms": 281973332.8471675, + "traffic_rms": 2.044278410582729 + }, + { + "batch_loss": 230.920654296875, + "innovation_rms": 4678648207.37208, + "instruction_rms": 4678648207.37208, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37119616.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.056667095615406e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2932025.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809.628662109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2355.320556640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2327409.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 249864.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271458.25, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 280909.78125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410031.71875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2932025.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809.53125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2355.320556640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2327409.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4678648207.37208, + "step": 40, + "teaching_rms": 4678648207.37208, + "traffic_rms": 1.2121148223786067 + }, + { + "batch_loss": 31762.796875, + "innovation_rms": 7326292460.262199, + "instruction_rms": 7326292460.262199, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33407654.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.510004204136038e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2638590.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2020.88232421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2112.53662109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2591268.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 274295.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 244312.421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 317812.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 369028.53125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2638590.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2020.784912109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2112.53662109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2591268.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7326292460.262199, + "step": 41, + "teaching_rms": 7326292460.262199, + "traffic_rms": 0.5155506315404988 + }, + { + "batch_loss": 12280.7861328125, + "innovation_rms": 10533069480.455685, + "instruction_rms": 10533069480.455685, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 30066888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.559003955521126e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2374472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2210.712890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1898.306640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2828715.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 296283.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219881.171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 351025.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 332125.65625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2374472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2210.615478515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1898.306640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2828715.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10533069480.455685, + "step": 42, + "teaching_rms": 10533069480.455685, + "traffic_rms": 0.3814607712067667 + }, + { + "batch_loss": 15.086641311645508, + "innovation_rms": 2050028.91916594, + "instruction_rms": 2050028.9191658907, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 27060200.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.703103731767706e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2136742.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2381.538330078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1708.2548828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3042389.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 316073.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197893.046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 380916.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 298913.09375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2136742.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2381.44091796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1708.2548828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3042389.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2050028.9191659065, + "step": 43, + "teaching_rms": 2050028.91916594, + "traffic_rms": 0.25354018047735316 + }, + { + "batch_loss": 21681.708984375, + "innovation_rms": 148828236.5599722, + "instruction_rms": 148828236.55997217, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24354180.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.932793186792243e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1922763.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2535.257568359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1537.191162109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3234666.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 333883.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178103.734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 407818.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 269021.78125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1922763.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2535.16015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1537.191162109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3234666.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 148828236.55997217, + "step": 44, + "teaching_rms": 148828236.5599722, + "traffic_rms": 0.42293929593634483 + }, + { + "batch_loss": 2436.09716796875, + "innovation_rms": 57932495.98792688, + "instruction_rms": 57932495.98792688, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21918762.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.239513739264e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730163.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2673.579345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1383.218505859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3407682.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 349912.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160293.359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 432030.65625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242119.59375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730163.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2673.48193359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1383.218505859375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3407682.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 57932495.98792688, + "step": 45, + "teaching_rms": 57932495.98792688, + "traffic_rms": 0.23322496978847904 + }, + { + "batch_loss": 12.205883979797363, + "innovation_rms": 185332.17738824905, + "instruction_rms": 185332.1773797239, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19726886.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.6155623653376e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1556806.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2798.042236328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1244.6292724609375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3563363.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 364339.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144264.015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 453821.40625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 217907.625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1556806.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2797.94482421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1244.6292724609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3563363.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 185332.17737990708, + "step": 46, + "teaching_rms": 185332.17738824905, + "traffic_rms": 0.26055376809742753 + }, + { + "batch_loss": 1941703.125, + "innovation_rms": 268195904690.025, + "instruction_rms": 268195904690.025, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17754198.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.05400612880384e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1400769.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2879.638671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1992.3671875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3703440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 377323.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 129837.609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 473433.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 196116.859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1400769.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2879.541259765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1992.367431640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3703440.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 268195904690.025, + "step": 47, + "teaching_rms": 268195904690.025, + "traffic_rms": 0.553397681970421 + }, + { + "batch_loss": 912708.125, + "innovation_rms": 29904884219.747295, + "instruction_rms": 29904884219.747295, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15978778.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.548605301175091e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1646047.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2939.28125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5987.8271484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3829472.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 389008.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116853.84375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 491083.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176505.171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1646047.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2939.183837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5987.8271484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3829472.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 29904884219.747295, + "step": 48, + "teaching_rms": 29904884219.747295, + "traffic_rms": 1.9042874416897249 + }, + { + "batch_loss": 1255296.125, + "innovation_rms": 20438229.777672652, + "instruction_rms": 20438229.77766848, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 74181400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.093744642208563e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1481521.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2991.13134765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8888.6943359375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3942863.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 399525.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105168.453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 506969.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158854.65625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1481521.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2991.033935546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8888.6943359375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3942863.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20438229.7776686, + "step": 49, + "teaching_rms": 20438229.777672652, + "traffic_rms": 2.2072521350169505 + }, + { + "batch_loss": 1903974.625, + "innovation_rms": 19851579.057827305, + "instruction_rms": 19851579.057822537, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 169664656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.267239229141811e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1333415.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3037.7666015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7763.541015625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4044875.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 408990.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94651.6015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 521266.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142969.1875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1333415.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3037.669189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7763.541015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4044875.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19851579.05782267, + "step": 50, + "teaching_rms": 19851579.057827305, + "traffic_rms": 2.3741489900013244 + }, + { + "batch_loss": 2190463.25, + "innovation_rms": 2387762127.413154, + "instruction_rms": 2387762127.413154, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 267703856.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.832115959404954e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1200052.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3079.7080078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6788.5595703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4136645.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 417509.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85186.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 534133.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128672.265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1200052.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3079.610595703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6788.5595703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4136645.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2387762127.413154, + "step": 51, + "teaching_rms": 2387762127.413154, + "traffic_rms": 1.5625218729893113 + }, + { + "batch_loss": 4965542.0, + "innovation_rms": 2545415669.384107, + "instruction_rms": 2545415669.384107, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 401632576.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8820543098165658e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1080087.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3117.42431640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5521.22314453125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4219198.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 425175.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83840.453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 545713.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115805.0390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1080087.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3117.326904296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5521.22314453125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4219198.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2545415669.384107, + "step": 52, + "teaching_rms": 2545415669.384107, + "traffic_rms": 2.1072715659253016 + }, + { + "batch_loss": 202927.78125, + "innovation_rms": 809458919.2205951, + "instruction_rms": 809458919.220595, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 476482848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.380121051290665e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 972062.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3165.904296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4959.1953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4293452.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 432075.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91619.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 556136.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104224.53125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 972062.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3165.89697265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4959.1953125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4293452.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 809458919.220595, + "step": 53, + "teaching_rms": 809458919.2205951, + "traffic_rms": 1.3654625140803247 + }, + { + "batch_loss": 437543.1875, + "innovation_rms": 1181350975.3742394, + "instruction_rms": 1181350975.3742394, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 511883136.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.110760271843164e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 874801.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3583.695068359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4437.04052734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4360239.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 438286.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93253.3046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 565516.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93802.078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 874801.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3583.687744140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4437.04052734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4360239.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1181350975.3742394, + "step": 54, + "teaching_rms": 1181350975.3742394, + "traffic_rms": 1.1014219835801182 + }, + { + "batch_loss": 4365.48583984375, + "innovation_rms": 2826992.510787667, + "instruction_rms": 2826992.510784653, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 505791648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.962044764449669e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 787235.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3959.6708984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3993.06689453125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4420303.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 443875.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83927.96875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 573958.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84421.8671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 787235.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3959.66357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3993.06689453125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4420303.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2826992.5107848444, + "step": 55, + "teaching_rms": 2826992.510787667, + "traffic_rms": 1.04184035953934 + }, + { + "batch_loss": 52128.76953125, + "innovation_rms": 1337658588.4103713, + "instruction_rms": 1337658588.4103713, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 668493440.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5616297678604861e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 708418.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4298.00927734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3593.450439453125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4474316.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 448905.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75535.171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 581556.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75979.6796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 708418.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4298.001953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3593.450439453125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4474316.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1337658588.4103713, + "step": 56, + "teaching_rms": 1337658588.4103713, + "traffic_rms": 0.6701106232185092 + }, + { + "batch_loss": 153.29733276367188, + "innovation_rms": 61663161.01280958, + "instruction_rms": 61663161.01280958, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 647877568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.475234817074266e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 637475.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4602.47119140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3233.759765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4522884.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 453432.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67981.65625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 588394.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68381.7109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 637475.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4602.4638671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3233.759765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4522884.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 61663161.01280958, + "step": 57, + "teaching_rms": 61663161.01280958, + "traffic_rms": 0.23537244363636223 + }, + { + "batch_loss": 4344.93310546875, + "innovation_rms": 1239857566566.8445, + "instruction_rms": 1239857566566.8445, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 626219648.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3913473025551892e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 573620.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4876.44091796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2910.006103515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4566549.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 457506.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61183.48828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 594549.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61543.5390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 573620.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4876.43359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2910.006103515625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4566549.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1239857566566.8445, + "step": 58, + "teaching_rms": 1239857566566.8445, + "traffic_rms": 0.21007295950872582 + }, + { + "batch_loss": 9596.78125, + "innovation_rms": 26984700321.19167, + "instruction_rms": 26984700321.19167, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 612894400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.296182097270014e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 516145.78125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5122.96484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2618.5986328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4605802.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 461173.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55065.13671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 600088.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55389.18359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 516145.78125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5122.95751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2618.5986328125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4605802.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26984700321.19167, + "step": 59, + "teaching_rms": 26984700321.19167, + "traffic_rms": 0.9950806862960249 + }, + { + "batch_loss": 10506.9619140625, + "innovation_rms": 3109262540.259982, + "instruction_rms": 3109262540.259982, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 631158848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2475227978659267e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 464413.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5344.78515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2356.3056640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4641084.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 464474.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49558.62109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 605073.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49850.265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 464413.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5344.77783203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2356.3056640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4641084.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3109262540.259982, + "step": 60, + "teaching_rms": 3109262540.259982, + "traffic_rms": 1.542005736222716 + }, + { + "batch_loss": 99652.6953125, + "innovation_rms": 19326434.329335883, + "instruction_rms": 19326434.32933566, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 734490368.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2308502158591918e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 417849.03125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5544.3701171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2120.218505859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4672791.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 467444.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44602.7578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 609559.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44865.23828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 417849.03125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5544.36279296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2120.218505859375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4672791.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19326434.32933571, + "step": 61, + "teaching_rms": 19326434.329335883, + "traffic_rms": 1.3815224269608173 + }, + { + "batch_loss": 39307.4453125, + "innovation_rms": 423187557105.0172, + "instruction_rms": 423187557105.0172, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 698487424.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1082091083490918e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 375937.21875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5723.94091796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1907.7188720703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4701281.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 470117.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40306.16015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 613597.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40378.71484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 375937.21875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5723.93359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1907.7188720703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4701281.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 423187557105.0172, + "step": 62, + "teaching_rms": 423187557105.0172, + "traffic_rms": 0.6199632296168288 + }, + { + "batch_loss": 94481.1796875, + "innovation_rms": 692888275090.4203, + "instruction_rms": 692888275090.4203, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 796846400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7052104558462894e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119389248.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5885.49755859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1716.4500732421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11938925.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 472523.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36275.54296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 617231.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36340.84375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119389248.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5885.490234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1716.4500732421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11938925.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 692888275090.4203, + "step": 63, + "teaching_rms": 692888275090.4203, + "traffic_rms": 1.9122632409068685 + }, + { + "batch_loss": 4687.9599609375, + "innovation_rms": 16039647972.605211, + "instruction_rms": 16039647972.605211, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 718932672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5346903585904394e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107449128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6030.83935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1544.291015625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22683838.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 474688.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32647.98828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 620502.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32706.7578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107449128.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6030.83203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1544.291015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22683838.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16039647972.605211, + "step": 64, + "teaching_rms": 16039647972.605211, + "traffic_rms": 0.27351372615930647 + }, + { + "batch_loss": 13.232213973999023, + "innovation_rms": 2.3971254569973466, + "instruction_rms": 2.396423775497152, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 647039424.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3812213502191862e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 96701944.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6161.5869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1389.3323974609375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32354032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476637.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29383.189453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 623445.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29436.08203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 96701944.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6161.57958984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1389.3323974609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32354032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.396423775497152, + "step": 65, + "teaching_rms": 2.3971254569973466, + "traffic_rms": 0.00020613807818989675 + }, + { + "batch_loss": 11.577930450439453, + "innovation_rms": 2.7198909224812495, + "instruction_rms": 2.7192547182358227, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 582335488.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2430991877094769e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87028512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6279.1982421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1249.8558349609375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41056884.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478391.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26444.869140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 626095.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26492.47265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87028512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6279.19091796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1249.8558349609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41056884.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.7192546572424914, + "step": 66, + "teaching_rms": 2.7198909224812495, + "traffic_rms": 0.00017788072686848792 + }, + { + "batch_loss": 7.672735214233398, + "innovation_rms": 2.7935484801303647, + "instruction_rms": 2.793064995172039, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 524101952.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1187892964263199e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78321552.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6384.9853515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1124.3143310546875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48889040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 479970.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23800.380859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 628479.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23843.224609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78321552.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6384.97802734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1124.3143310546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48889040.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.7930649357905333, + "step": 67, + "teaching_rms": 2.7935484801303647, + "traffic_rms": 0.00013493305201180085 + }, + { + "batch_loss": 5.158134460449219, + "innovation_rms": 2.797273445401349, + "instruction_rms": 2.7968937429306924, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 471691744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0069103461678449e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70484504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6480.1298828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1011.3157958984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55937492.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 481390.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21420.341796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 630625.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21458.90234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70484504.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6480.12255859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1011.3157958984375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55937492.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.7968937429306924, + "step": 68, + "teaching_rms": 2.797273445401349, + "traffic_rms": 9.7826113813053e-05 + }, + { + "batch_loss": 3.546567916870117, + "innovation_rms": 3.2626760591682067, + "instruction_rms": 3.262585249390722, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 424522560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.062193390388511e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63430460.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6565.6953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 909.6069946289062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 62280540.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 482669.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19278.306640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 632556.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19313.01171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63430460.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6565.68798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 909.6069946289062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 62280540.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.2625852239727804, + "step": 69, + "teaching_rms": 3.2626760591682067, + "traffic_rms": 8.492383034989884e-05 + }, + { + "batch_loss": 3.5565946102142334, + "innovation_rms": 1.503123520278364, + "instruction_rms": 1.5029966337501435, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 382070304.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15597384519123e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57081184.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6642.638671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.0599975585938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67988656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 483820.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17350.474609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 634294.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17381.7109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57081184.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6642.63134765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.0599975585938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67988656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.5029966337501435, + "step": 70, + "teaching_rms": 1.503123520278364, + "traffic_rms": 8.904332346617729e-05 + }, + { + "batch_loss": 3.8751304149627686, + "innovation_rms": 4.818269196306638, + "instruction_rms": 4.818209000872502, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 343863264.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.340376735550013e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51366264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6711.8212890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 735.6594848632812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 73125280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 484855.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15615.4267578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 635859.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15643.5390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51366264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6711.81396484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 735.6594848632812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 73125280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.818209000872502, + "step": 71, + "teaching_rms": 4.818269196306638, + "traffic_rms": 9.884947512599928e-05 + }, + { + "batch_loss": 4.165862560272217, + "innovation_rms": 5.08922027637132, + "instruction_rms": 5.089129896078254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 309476928.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.606339336872919e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46222324.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6774.0185546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 661.4916381835938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 77747512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 485787.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14053.8837890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 637267.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14079.1845703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46222324.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6774.01123046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 661.4916381835938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 77747512.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.089129896078254, + "step": 72, + "teaching_rms": 5.08922027637132, + "traffic_rms": 0.00011512788364478431 + }, + { + "batch_loss": 4.397496700286865, + "innovation_rms": 4.794923112798185, + "instruction_rms": 4.794807121909659, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 278529248.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.94570533446615e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41592316.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6829.92822265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 594.7340087890625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 81906744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 486626.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12648.4951171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 638534.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12671.265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41592316.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6829.9208984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 594.7340087890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 81906744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.794807121909659, + "step": 73, + "teaching_rms": 4.794923112798185, + "traffic_rms": 0.00013768399429907403 + }, + { + "batch_loss": 4.935162544250488, + "innovation_rms": 5.145020565374717, + "instruction_rms": 5.1448832957463155, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 250676320.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.3511348010195354e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37424892.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6880.1787109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 534.646240234375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 85649232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 487381.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11383.6455078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 639674.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11404.138671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37424892.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6880.17138671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 534.646240234375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 85649232.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.144883279627738, + "step": 74, + "teaching_rms": 5.145020565374717, + "traffic_rms": 0.00016276549668208488 + }, + { + "batch_loss": 5.190221786499023, + "innovation_rms": 4.625468580170603, + "instruction_rms": 4.625262875298332, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 225608688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.8160213896370586e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33673836.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6925.33544921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 480.5618591308594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 89016616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 488061.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10245.2802734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 640700.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10263.724609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33673836.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6925.328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 480.5618591308594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 89016616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.625262857368929, + "step": 75, + "teaching_rms": 4.625468580170603, + "traffic_rms": 0.00018383958000573654 + }, + { + "batch_loss": 4.133368015289307, + "innovation_rms": 4.8530575648461065, + "instruction_rms": 4.852895025083289, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 203047824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3344191475941376e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30297550.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6965.9072265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 431.8811340332031, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92046368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 488672.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9220.751953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 641624.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9237.3515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30297550.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6965.89990234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 431.8811340332031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 92046368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.852895025083289, + "step": 76, + "teaching_rms": 4.8530575648461065, + "traffic_rms": 0.00019617551664209915 + }, + { + "batch_loss": 3.521427631378174, + "innovation_rms": 4.706128230806228, + "instruction_rms": 4.706021923555655, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 182743040.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.900977232834724e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27258590.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7002.35205078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 388.06414794921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 94772224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 489223.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8298.6767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 642456.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8313.6162109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27258590.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7002.3447265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 388.06414794921875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 94772224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.706021923555655, + "step": 77, + "teaching_rms": 4.706128230806228, + "traffic_rms": 0.00020049072800498057 + }, + { + "batch_loss": 3.371114492416382, + "innovation_rms": 4.672610795498748, + "instruction_rms": 4.67248423816426, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 164468736.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.510879440831775e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24523252.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7035.08251953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 348.6249694824219, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97224552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 489718.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7468.80908203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 643204.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7482.25439453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24523252.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7035.0751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 348.6249694824219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97224552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.67248423816426, + "step": 78, + "teaching_rms": 4.672610795498748, + "traffic_rms": 0.00019796900818293644 + }, + { + "batch_loss": 3.109663963317871, + "innovation_rms": 4.725532616192604, + "instruction_rms": 4.725399328658857, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 148021856.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.159791634187551e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22061204.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7064.46923828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 313.126220703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 99430672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 490164.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6721.92822265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 643877.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6734.02880859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22061204.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7064.4619140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 313.126220703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 99430672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.725399328658857, + "step": 79, + "teaching_rms": 4.725532616192604, + "traffic_rms": 0.00018957258177290174 + }, + { + "batch_loss": 2.9990968704223633, + "innovation_rms": 4.901509493289887, + "instruction_rms": 4.901408614918981, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 133219672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8438124364090573e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19845140.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7090.8466796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 281.1742248535156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101415184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 490565.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6049.7353515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 644483.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6060.6259765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19845140.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7090.83935546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 281.1742248535156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101415184.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.901408614918981, + "step": 80, + "teaching_rms": 4.901509493289887, + "traffic_rms": 0.00017728751118882436 + }, + { + "batch_loss": 2.750286817550659, + "innovation_rms": 4.9985145025468745, + "instruction_rms": 4.998440842672328, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 119897704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.559431141228544e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17850484.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7114.515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 252.41461181640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103200232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 490926.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5444.76171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 645029.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5454.5634765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17850484.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7114.50830078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 252.41461181640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103200232.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.998440842672328, + "step": 81, + "teaching_rms": 4.9985145025468745, + "traffic_rms": 0.00016400621600418385 + }, + { + "batch_loss": 2.6851541996002197, + "innovation_rms": 4.720001240682983, + "instruction_rms": 4.719937130050532, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 107907936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3034880271056896e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16055115.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7135.74658203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 226.5284423828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104805744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 491251.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4900.28564453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 645520.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4909.10693359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16055115.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7135.7392578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 226.52842712402344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104805744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.719937130050532, + "step": 82, + "teaching_rms": 4.720001240682983, + "traffic_rms": 0.00015421784302436739 + }, + { + "batch_loss": 2.706390857696533, + "innovation_rms": 4.866454057057827, + "instruction_rms": 4.866370540172784, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 97117144.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0731391384957747e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14439122.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7154.783203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 203.22862243652344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106249656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 491544.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4410.2568359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 645961.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4418.1962890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14439122.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7154.77587890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 203.22860717773438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106249656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.866370540172784, + "step": 83, + "teaching_rms": 4.866454057057827, + "traffic_rms": 0.00015072496009595622 + }, + { + "batch_loss": 2.666552782058716, + "innovation_rms": 5.162129767890629, + "instruction_rms": 5.161987897815347, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 87405432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.865825190286459e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12984584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7171.84423828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.2567596435547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107548112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 491807.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3969.23095703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 646359.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3976.37646484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12984584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7171.8369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.25674438476562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107548112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.161987897815347, + "step": 84, + "teaching_rms": 5.162129767890629, + "traffic_rms": 0.00015333141253582707 + }, + { + "batch_loss": 2.8221521377563477, + "innovation_rms": 5.368359635439227, + "instruction_rms": 5.368266567990865, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 78664888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6792427056175514e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11675370.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7187.12744140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.38026428222656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108715648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492044.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3572.307861328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 646717.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3578.73876953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11675370.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7187.1201171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.38023376464844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108715648.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.36826658343872, + "step": 85, + "teaching_rms": 5.368359635439227, + "traffic_rms": 0.00015983301917612104 + }, + { + "batch_loss": 2.821532964706421, + "innovation_rms": 5.485180214611453, + "instruction_rms": 5.485133638195781, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 70798400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.511318417875927e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10496961.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7200.810546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.3897705078125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109765344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492257.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3215.076904296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 647039.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3220.86474609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10496961.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7200.80322265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.38973999023438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109765344.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.4851336533145005, + "step": 86, + "teaching_rms": 5.485180214611453, + "traffic_rms": 0.0001678555803921571 + }, + { + "batch_loss": 2.752714157104492, + "innovation_rms": 5.556313132344045, + "instruction_rms": 5.556257972549476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 63718560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3601865932682035e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9436288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7213.05322265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 131.0968780517578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110708976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492449.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2893.569091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 647329.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2898.778076171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9436288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7213.0458984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 131.09683227539062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110708976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.556257972549476, + "step": 87, + "teaching_rms": 5.556313132344045, + "traffic_rms": 0.00017610965227515157 + }, + { + "batch_loss": 2.748995542526245, + "innovation_rms": 6.087034774543276, + "instruction_rms": 6.0869600772458154, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 57346704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2241678995816448e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8481588.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7223.99951171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.33195495605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111557136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492622.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2604.212158203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 647590.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2608.900146484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8481588.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7223.9921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.3319091796875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111557136.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.0869600772458154, + "step": 88, + "teaching_rms": 6.087034774543276, + "traffic_rms": 0.00018318586203247268 + }, + { + "batch_loss": 2.6797497272491455, + "innovation_rms": 6.16325770730061, + "instruction_rms": 6.163264795842611, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 51612032.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1017511096234803e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7622273.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7233.779296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.94234466552734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112319360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492777.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2343.790771484375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 647825.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2348.010009765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7622273.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7233.77197265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.94230651855469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112319360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.163264795842611, + "step": 89, + "teaching_rms": 6.16325770730061, + "traffic_rms": 0.00018791090047872813 + }, + { + "batch_loss": 2.7976326942443848, + "innovation_rms": 6.3621287554766885, + "instruction_rms": 6.3621369035269435, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 46450828.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.915760072510669e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6848814.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7242.5087890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.7906494140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113004240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492917.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2109.41162109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648036.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2113.208984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6848814.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7242.50146484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.79061126708984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113004240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.3621369035269435, + "step": 90, + "teaching_rms": 6.3621287554766885, + "traffic_rms": 0.0001903813976799624 + }, + { + "batch_loss": 2.6023120880126953, + "innovation_rms": 6.6915586080730405, + "instruction_rms": 6.691599414354479, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 41805744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.924184408856986e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6152632.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7250.29248046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.75318908691406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113619504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493043.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1898.470458984375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648226.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1901.8880615234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6152632.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7250.28515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.75314331054688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113619504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.691599414354479, + "step": 91, + "teaching_rms": 6.6915586080730405, + "traffic_rms": 0.00019004047501469939 + }, + { + "batch_loss": 2.450298309326172, + "innovation_rms": 6.994762748468965, + "instruction_rms": 6.994752012584559, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37625168.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.031765796172595e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5526006.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7257.2255859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.71863555908203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 114172104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493156.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1708.6234130859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648397.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1711.69921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5526006.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7257.21826171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.71859741210938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 114172104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.994752024440333, + "step": 92, + "teaching_rms": 6.994762748468965, + "traffic_rms": 0.0001880583912234487 + }, + { + "batch_loss": 2.5072805881500244, + "innovation_rms": 6.704280970986329, + "instruction_rms": 6.70431064102478, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33862652.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.228588872957952e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4961988.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7263.392578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.5867919921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 114668304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493258.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1537.760986328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648551.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1540.529296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4961988.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7263.38525390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.58675384521484, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 114668304.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.70431064102478, + "step": 93, + "teaching_rms": 6.704280970986329, + "traffic_rms": 0.00018514130613491272 + }, + { + "batch_loss": 2.421393632888794, + "innovation_rms": 6.8431558703500635, + "instruction_rms": 6.843146243033615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 30476386.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.505729985662157e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4454322.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7268.87060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.2674674987793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115113736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493350.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1383.98486328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648690.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1386.476318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4454322.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7268.86328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.26742935180664, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115113736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.843146243033615, + "step": 94, + "teaching_rms": 6.8431558703500635, + "traffic_rms": 0.00018168454083461114 + }, + { + "batch_loss": 2.3856616020202637, + "innovation_rms": 6.536769646642458, + "instruction_rms": 6.536758632540967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 27428748.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.85515711594496e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3997379.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7273.72802734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.679481506347656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115513472.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493433.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1245.5863037109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648815.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1247.82861328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3997379.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7273.720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.679447174072266, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115513472.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.536758632540967, + "step": 95, + "teaching_rms": 6.536769646642458, + "traffic_rms": 0.00017827013504081383 + }, + { + "batch_loss": 2.4668610095977783, + "innovation_rms": 6.680495957511319, + "instruction_rms": 6.6804750013567284, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24685874.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.269641404350464e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3586089.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7278.02685546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.7497673034668, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115872080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493507.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1121.027587890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 648927.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1123.0457763671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3586089.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7278.01953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.749732971191406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115872080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.6804750013567284, + "step": 96, + "teaching_rms": 6.680495957511319, + "traffic_rms": 0.0001754678699377362 + }, + { + "batch_loss": 2.4082882404327393, + "innovation_rms": 6.942897179607749, + "instruction_rms": 6.94288501476497, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 22217286.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.742677263915418e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3215893.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281.8232421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.412559509277344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116193672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493574.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1008.9248046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649028.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010.7411499023438, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3215893.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281.81591796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.41252517700195, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116193672.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.942885026709313, + "step": 97, + "teaching_rms": 6.942897179607749, + "traffic_rms": 0.00017340504889759855 + }, + { + "batch_loss": 2.3376383781433105, + "innovation_rms": 7.500890321570429, + "instruction_rms": 7.500868070158768, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19995558.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.268409365725184e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2882684.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7285.1669921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.6086540222168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116481944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493634.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 908.0322875976562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649119.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 909.6669921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2882684.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7285.15966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.608619689941406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116481944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.500868070158768, + "step": 98, + "teaching_rms": 7.500890321570429, + "traffic_rms": 0.00017294201591820953 + }, + { + "batch_loss": 2.3640804290771484, + "innovation_rms": 7.191298602043249, + "instruction_rms": 7.191295620313056, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17996002.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.841568214404301e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2582768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7288.103515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.28477478027344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116740224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493689.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 817.2290649414062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649201.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.7002563476562, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2582768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7288.09619140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.28474044799805, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116740224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.191295631844803, + "step": 99, + "teaching_rms": 7.191298602043249, + "traffic_rms": 0.0001743959792387359 + }, + { + "batch_loss": 2.441969156265259, + "innovation_rms": 7.6006923661825745, + "instruction_rms": 7.600699058491957, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 16196402.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.457411307064525e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2312817.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7290.67333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.392961502075195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116971504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493737.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 735.5061645507812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649275.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 736.8302001953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2312817.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7290.666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.392927169799805, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116971504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.600699069402558, + "step": 100, + "teaching_rms": 7.6006923661825745, + "traffic_rms": 0.0001773259545533088 + }, + { + "batch_loss": 2.40368390083313, + "innovation_rms": 7.346134682714178, + "instruction_rms": 7.346139891186391, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14576762.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1116700904587264e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2069838.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7292.91357421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.890045166015625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117178488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493781.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 661.9555053710938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649341.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 663.1471557617188, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2069838.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7292.90625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.890012741088867, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117178488.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.346139891186391, + "step": 101, + "teaching_rms": 7.346134682714178, + "traffic_rms": 0.0001812375067321766 + }, + { + "batch_loss": 2.3088324069976807, + "innovation_rms": 7.514404986027673, + "instruction_rms": 7.514391556442994, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13119086.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.800503145837363e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851136.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294.85693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.737171173095703, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117363600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493821.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 595.7599487304688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649401.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 596.8323974609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851136.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294.849609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.737138748168945, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117363600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.514391556442994, + "step": 102, + "teaching_rms": 7.514404986027673, + "traffic_rms": 0.00018529983783960432 + }, + { + "batch_loss": 2.373708963394165, + "innovation_rms": 6.9128208891735135, + "instruction_rms": 6.912761439395106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11807177.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5204528097787904e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1654286.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7296.53271484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.899368286132812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117529032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493856.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 536.1839599609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649454.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 537.149169921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1654286.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7296.525390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.899335861206055, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117529032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.912761439395106, + "step": 103, + "teaching_rms": 6.9128208891735135, + "traffic_rms": 0.00018879767070240428 + }, + { + "batch_loss": 2.321204900741577, + "innovation_rms": 6.83821128938339, + "instruction_rms": 6.838138529163291, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10626459.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.268407593225421e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1477104.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7297.96826171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.345157623291016, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117676744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493888.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 482.5655517578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649503.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 483.4342346191406, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1477104.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7297.9609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.345123291015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117676744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.838138529163291, + "step": 104, + "teaching_rms": 6.83821128938339, + "traffic_rms": 0.00019065794482234138 + }, + { + "batch_loss": 2.306959629058838, + "innovation_rms": 7.05702767264259, + "instruction_rms": 7.056979449062498, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9563813.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.041566748003533e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1317626.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7299.18701171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.04620361328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117808504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493917.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 434.3089904785156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649546.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 435.0907897949219, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1317626.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7299.1796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.04616928100586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117808504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.05697946081373, + "step": 105, + "teaching_rms": 7.05702767264259, + "traffic_rms": 0.00019073670972054752 + }, + { + "batch_loss": 2.3066723346710205, + "innovation_rms": 6.893535710821329, + "instruction_rms": 6.893492043706694, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8607432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8374099873038336e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1174083.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7300.2109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.977004051208496, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117925912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493943.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 390.8780822753906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649585.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 391.5816955566406, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1174083.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7300.20361328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.976970672607422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117925912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.893492043706694, + "step": 106, + "teaching_rms": 6.893535710821329, + "traffic_rms": 0.00018946108937008672 + }, + { + "batch_loss": 2.368769884109497, + "innovation_rms": 6.629430440052459, + "instruction_rms": 6.629353349833633, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7746689.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6536690315231232e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1044882.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.0595703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.114605903625488, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118030400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493966.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.7902526855469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649621.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 352.42352294921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1044882.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.05224609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.114571571350098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118030400.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.629353374852122, + "step": 107, + "teaching_rms": 6.629430440052459, + "traffic_rms": 0.00018741543954835425 + }, + { + "batch_loss": 2.3642804622650146, + "innovation_rms": 6.852547648854232, + "instruction_rms": 6.852467440052597, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6972020.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4883021605830656e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 928590.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.75048828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.438345909118652, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118123256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 493987.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 316.6112060546875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649652.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 317.18115234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 928590.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.7431640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.438311576843262, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118123256.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.852467452154544, + "step": 108, + "teaching_rms": 6.852547648854232, + "traffic_rms": 0.00018466704916437834 + }, + { + "batch_loss": 2.3419060707092285, + "innovation_rms": 6.882066668345699, + "instruction_rms": 6.88201079406638, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6274818.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.339471987474432e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 823919.4375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.29931640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.957943439483643, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118205648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494006.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 284.9500732421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649681.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.4630432128906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 823919.4375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.2919921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.958189964294434, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118205648.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.88201079406638, + "step": 109, + "teaching_rms": 6.882066668345699, + "traffic_rms": 0.000181859255250394 + }, + { + "batch_loss": 2.4305567741394043, + "innovation_rms": 6.991262565580623, + "instruction_rms": 6.99119507588205, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5647336.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2055247350398976e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 729706.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.72021484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.17030143737793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118278616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494023.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256.4550476074219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649707.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256.9167175292969, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 729706.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.712890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.170547962188721, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118278616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.991195087743857, + "step": 110, + "teaching_rms": 6.991262565580623, + "traffic_rms": 0.00017972490214511935 + }, + { + "batch_loss": 2.3168745040893555, + "innovation_rms": 6.153435853351742, + "instruction_rms": 6.153358890634045, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5082602.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0849723044855808e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 644908.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.02587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.461362361907959, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118343104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494039.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.80953979492188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649730.125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 231.22503662109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 644908.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.0185546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.46160888671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118343104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.153358890634045, + "step": 111, + "teaching_rms": 6.153435853351742, + "traffic_rms": 0.00017837359625423726 + }, + { + "batch_loss": 2.355905055999756, + "innovation_rms": 6.830606561194431, + "instruction_rms": 6.830547243833888, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4574342.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9764750955118592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568583.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.22802734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.8232622146606445, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118399960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494052.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.72857666015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649750.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.10252380371094, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568583.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.220703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.8235087394714355, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118399960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.830547243833888, + "step": 112, + "teaching_rms": 6.830606561194431, + "traffic_rms": 0.00017805420459192307 + }, + { + "batch_loss": 2.3343265056610107, + "innovation_rms": 6.599697540372068, + "instruction_rms": 6.599659165166549, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4116907.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8788275644858368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 499884.84375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.3369140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2489237785339355, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118449952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494065.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.95571899414062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649769.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.29226684570312, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 499884.84375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.32958984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.249170780181885, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118449952.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.599659165166549, + "step": 113, + "teaching_rms": 6.599697540372068, + "traffic_rms": 0.0001788067401599778 + }, + { + "batch_loss": 2.304056167602539, + "innovation_rms": 6.45386085379656, + "instruction_rms": 6.45382261425515, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3705217.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7909448026685440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438051.34375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.36181640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7319769859313965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118493760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494076.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.26014709472656, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649786.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.5630340576172, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438051.34375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.3544921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7322239875793457, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118493760.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.4538226271046195, + "step": 114, + "teaching_rms": 6.45386085379656, + "traffic_rms": 0.00018055676024407628 + }, + { + "batch_loss": 2.323514461517334, + "innovation_rms": 6.964082438903883, + "instruction_rms": 6.96405710943639, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3334695.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7118502955581440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 382396.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.31103515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2666876316070557, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118532000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494086.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.4341278076172, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649801.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.70672607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 382396.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.3037109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.266934394836426, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118532000.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.96405710943639, + "step": 115, + "teaching_rms": 6.964082438903883, + "traffic_rms": 0.00018291254960408577 + }, + { + "batch_loss": 2.335228204727173, + "innovation_rms": 6.799491662180041, + "instruction_rms": 6.799446500820315, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3001225.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6406652391587840.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 332303.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.1923828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9924120903015137, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118565232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494095.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.29071044921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649815.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.5360565185547, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 332303.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.18505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9923160076141357, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118565232.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.799446500820315, + "step": 116, + "teaching_rms": 6.799491662180041, + "traffic_rms": 0.00018546060607765657 + }, + { + "batch_loss": 2.3172643184661865, + "innovation_rms": 6.4817772551770805, + "instruction_rms": 6.481738760175363, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2701103.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5765987152429056.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287217.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.0126953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0366623401641846, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118593952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494103.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.66163635253906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649827.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.8824462890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287217.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7303.00537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0365662574768066, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118593952.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.481738772969491, + "step": 117, + "teaching_rms": 6.4817772551770805, + "traffic_rms": 0.00018740214655085205 + }, + { + "batch_loss": 2.329347610473633, + "innovation_rms": 6.616700019648508, + "instruction_rms": 6.616623995431451, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2430993.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5189388329811968.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 246635.90625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.77783203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0764572620391846, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118618616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494110.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.39546966552734, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649838.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.59420013427734, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 246635.90625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.7705078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0763609409332275, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118618616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.616623995431451, + "step": 118, + "teaching_rms": 6.616700019648508, + "traffic_rms": 0.00018837874239663666 + }, + { + "batch_loss": 2.323272943496704, + "innovation_rms": 6.576141437976522, + "instruction_rms": 6.57612195582765, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2187893.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4670449443143680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210110.453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.49365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.112241744995117, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118639624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494117.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.35591888427734, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649848.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.53477478027344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210110.453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.486328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.11214542388916, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118639624.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.57612195582765, + "step": 119, + "teaching_rms": 6.576141437976522, + "traffic_rms": 0.00018814273045506699 + }, + { + "batch_loss": 2.3347325325012207, + "innovation_rms": 6.717809121372995, + "instruction_rms": 6.717763348868136, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1969104.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4203404498829312.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177235.4375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.16455078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.289669990539551, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118657344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494123.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.42032623291016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649857.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.58129119873047, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177235.4375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7302.1572265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.289661407470703, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118657344.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.717763348868136, + "step": 120, + "teaching_rms": 6.717809121372995, + "traffic_rms": 0.00018716177360587685 + }, + { + "batch_loss": 2.317718029022217, + "innovation_rms": 6.453124052987674, + "instruction_rms": 6.453116805915821, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1772193.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3783064102633472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147646.15625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.79541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6909193992614746, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118672112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494128.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.4782943725586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649865.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.62316131591797, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147646.15625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.7880859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.690911054611206, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118672112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.453116805915821, + "step": 121, + "teaching_rms": 6.453124052987674, + "traffic_rms": 0.00018557402918585524 + }, + { + "batch_loss": 2.3198153972625732, + "innovation_rms": 7.030523977118126, + "instruction_rms": 7.030505669518461, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1594974.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3404757746057216.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121014.3203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.39013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.052006721496582, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118684216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494133.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.43046569824219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649872.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.56084442138672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121014.3203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7301.3828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.051998615264893, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118684216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.030505669518461, + "step": 122, + "teaching_rms": 7.030523977118126, + "traffic_rms": 0.00018391121334532077 + }, + { + "batch_loss": 2.32063364982605, + "innovation_rms": 7.29919445251697, + "instruction_rms": 7.2991978030033255, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1435477.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3064281998295040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97044.4609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7300.95263671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3769450187683105, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118693920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494137.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.18741607666016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649879.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.30475616455078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97044.4609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7300.9453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.376936912536621, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118693920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.299197814364602, + "step": 123, + "teaching_rms": 7.29919445251697, + "traffic_rms": 0.00018258331859138938 + }, + { + "batch_loss": 2.3266499042510986, + "innovation_rms": 6.9046094494050365, + "instruction_rms": 6.904587939290267, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1291929.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2757853798465536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75470.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7300.48583984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.669345855712891, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118701464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494141.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.66867446899414, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649885.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.7742805480957, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75470.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7300.478515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.669337749481201, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118701464.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.904587939290267, + "step": 124, + "teaching_rms": 6.9046094494050365, + "traffic_rms": 0.00018203682280998733 + }, + { + "batch_loss": 2.3482472896575928, + "innovation_rms": 7.205223072185747, + "instruction_rms": 7.205221138361617, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1162736.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2482068311244800.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56053.4140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7299.99267578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.932459831237793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118707072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494145.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.80180740356445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649890.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.8968505859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56053.4140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7299.9853515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.9324517250061035, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118707072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.205221138361617, + "step": 125, + "teaching_rms": 7.205223072185747, + "traffic_rms": 0.0001820925668504731 + }, + { + "batch_loss": 2.299316167831421, + "innovation_rms": 7.06156314067772, + "instruction_rms": 7.061536658562976, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1046462.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2233861480120320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38577.36328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7299.4755859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.169212818145752, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118710928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494148.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.52162551879883, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649895.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.60716247558594, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38577.36328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7299.46826171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.169205188751221, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118710928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.061536670306625, + "step": 126, + "teaching_rms": 7.06156314067772, + "traffic_rms": 0.00018286788810060976 + }, + { + "batch_loss": 2.302696943283081, + "innovation_rms": 7.002768598332177, + "instruction_rms": 7.002735723704907, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 941816.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2010475264999424.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22848.53125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7298.9375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.382238864898682, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118713216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494151.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.76946258544922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649899.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.8464469909668, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22848.53125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7298.93017578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.38223123550415, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118713216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.002735723704907, + "step": 127, + "teaching_rms": 7.002768598332177, + "traffic_rms": 0.0001839908717402859 + }, + { + "batch_loss": 2.3045222759246826, + "innovation_rms": 7.097911601273733, + "instruction_rms": 7.0978818560826165, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 847634.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809427711655936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692.3564453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7298.3798828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.57390832901001, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118714088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494153.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.492515563964844, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649903.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.56180191040039, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692.3564453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7298.37255859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.573901176452637, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118714088.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.09788186776613, + "step": 128, + "teaching_rms": 7.097911601273733, + "traffic_rms": 0.00018511647531271758 + }, + { + "batch_loss": 2.3170344829559326, + "innovation_rms": 7.073894651574195, + "instruction_rms": 7.07384286060744, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 762871.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1628484900225024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4683.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7297.80517578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.746355056762695, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118713680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494156.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.64326477050781, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649907.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.70561981201172, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4683.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7297.7978515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.7463483810424805, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118713680.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.07384286060744, + "step": 129, + "teaching_rms": 7.073894651574195, + "traffic_rms": 0.00018599926804300802 + }, + { + "batch_loss": 2.306735038757324, + "innovation_rms": 7.1146445630258714, + "instruction_rms": 7.114593780284939, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 686584.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1465636383358976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15514.8271484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7297.21484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.901500225067139, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118712128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494158.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.178937911987305, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649910.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.235057830810547, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15514.8271484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7297.20751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.901493072509766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118712128.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.114593780284939, + "step": 130, + "teaching_rms": 7.1146445630258714, + "traffic_rms": 0.0001862780951649457 + }, + { + "batch_loss": 2.3144664764404297, + "innovation_rms": 6.419682306130442, + "instruction_rms": 6.419607838868094, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 617925.8125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1319072771866624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25834.556640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7296.61083984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.04107141494751, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118709544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494160.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.06104278564453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649913.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.11155128479004, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25834.556640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7296.603515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.041064262390137, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118709544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.419607838868094, + "step": 131, + "teaching_rms": 6.419682306130442, + "traffic_rms": 0.00018599312914720353 + }, + { + "batch_loss": 2.3052940368652344, + "innovation_rms": 7.08241734697993, + "instruction_rms": 7.082374025324468, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 556133.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1187165467836416.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35122.0546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7295.994140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.166625022888184, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118706032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494161.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.25493812561035, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649915.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.300395965576172, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35122.0546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7295.98681640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.166618347167969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118706032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.082374025324468, + "step": 132, + "teaching_rms": 7.08241734697993, + "traffic_rms": 0.000185015729464906 + }, + { + "batch_loss": 2.2620248794555664, + "innovation_rms": 7.446686604524191, + "instruction_rms": 7.4466545310844126, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 500519.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1068448947896320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43480.44921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7295.3662109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.279561996459961, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118701680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494163.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.72944450378418, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649917.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.770355224609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43480.44921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7295.35888671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.279554843902588, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118701680.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.4466545310844126, + "step": 133, + "teaching_rms": 7.446686604524191, + "traffic_rms": 0.0001838069631073833 + }, + { + "batch_loss": 2.371011734008789, + "innovation_rms": 7.2102318997984245, + "instruction_rms": 7.210179457540445, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 450467.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 961604053106688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51002.5703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294.72802734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.3811421394348145, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118696576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494164.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.456499099731445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649919.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.493318557739258, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51002.5703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294.720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.3811354637146, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118696576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.210179457540445, + "step": 134, + "teaching_rms": 7.2102318997984245, + "traffic_rms": 0.00018304860875225523 + }, + { + "batch_loss": 2.383371114730835, + "innovation_rms": 6.675887495918539, + "instruction_rms": 6.67584024858995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 405421.15625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 865443627663360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57771.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294.08056640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.472500324249268, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118690800.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494165.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.41084861755371, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649921.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.443986892700195, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57771.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7294.0732421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.472494125366211, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118690800.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.6758402610120875, + "step": 135, + "teaching_rms": 6.675887495918539, + "traffic_rms": 0.00018221247441891034 + }, + { + "batch_loss": 2.337975263595581, + "innovation_rms": 7.118761526160655, + "instruction_rms": 7.118720395054209, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 364879.03125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 778899231342592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63863.8515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7293.42529296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.554657936096191, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118684416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494166.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.56976318359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649923.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.59958839416504, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63863.8515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7293.41796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.554651737213135, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118684416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.118720395054209, + "step": 136, + "teaching_rms": 7.118761526160655, + "traffic_rms": 0.00018141183427241163 + }, + { + "batch_loss": 2.301238775253296, + "innovation_rms": 6.972193186541392, + "instruction_rms": 6.97216102311542, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 328391.125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 701009328340992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69345.90625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7292.76220703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.628534317016602, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118677480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494167.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.912786483764648, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649924.875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.939629554748535, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69345.90625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7292.7548828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.628528594970703, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118677480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.97216102311542, + "step": 137, + "teaching_rms": 6.972193186541392, + "traffic_rms": 0.000181246711164748 + }, + { + "batch_loss": 2.3048219680786133, + "innovation_rms": 7.438790543904323, + "instruction_rms": 7.438740688511563, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 295552.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 630908415639552.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74279.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7292.0927734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.6949567794799805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118670056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494168.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.421507835388184, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649926.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.445666313171387, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74279.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7292.08544921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.694951057434082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118670056.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.438740699659713, + "step": 138, + "teaching_rms": 7.438790543904323, + "traffic_rms": 0.00018192536499705603 + }, + { + "batch_loss": 2.31916880607605, + "innovation_rms": 6.646106939239022, + "instruction_rms": 6.6460664747967115, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 265996.8125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 567817560653824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78718.1640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7291.41748046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.754670143127441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118662184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494169.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.079357147216797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649927.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.101099014282227, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78718.1640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7291.41015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.754664421081543, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118662184.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.6460664747967115, + "step": 139, + "teaching_rms": 6.646106939239022, + "traffic_rms": 0.00018303443150087612 + }, + { + "batch_loss": 2.3096160888671875, + "innovation_rms": 7.357035523421746, + "instruction_rms": 7.357025715970476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 239397.125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 511035811299328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82712.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7290.73681640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.808344841003418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118653912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494170.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.871420860290527, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649928.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.89098834991455, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82712.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7290.7294921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.8083391189575195, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118653912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.357025715970476, + "step": 140, + "teaching_rms": 7.357035523421746, + "traffic_rms": 0.00018417299314360983 + }, + { + "batch_loss": 2.3113739490509033, + "innovation_rms": 7.3710232739502946, + "instruction_rms": 7.3710130239556575, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 215457.40625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 459932243591168.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86306.6953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7290.05126953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.856583595275879, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118645280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494171.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.784278869628906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649929.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.801889419555664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86306.6953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7290.0439453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.8565778732299805, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118645280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.3710130239556575, + "step": 141, + "teaching_rms": 7.3710232739502946, + "traffic_rms": 0.0001855635743541069 + }, + { + "batch_loss": 2.3135807514190674, + "innovation_rms": 6.6724355379330245, + "instruction_rms": 6.672372566187453, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 193911.671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 413939015876608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89540.546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7289.361328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.899930000305176, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118636328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494171.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.805850982666016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649930.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.821700096130371, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89540.546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7289.35400390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.899924278259277, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118636328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.672372566187453, + "step": 142, + "teaching_rms": 6.6724355379330245, + "traffic_rms": 0.0001869685455571913 + }, + { + "batch_loss": 2.327869415283203, + "innovation_rms": 7.214495935635864, + "instruction_rms": 7.214508902642965, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 174520.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 372545127710720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 92450.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7288.66748046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.938872814178467, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118627080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494172.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.925265789031982, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649931.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.939529895782471, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 92450.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7288.66015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.938867092132568, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118627080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.214508902642965, + "step": 143, + "teaching_rms": 7.214495935635864, + "traffic_rms": 0.00018756397214223856 + }, + { + "batch_loss": 2.3173410892486572, + "innovation_rms": 6.604074818469429, + "instruction_rms": 6.604036961720399, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 157068.453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 335290614939648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95067.8203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7287.97021484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.973852157592773, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118617576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494172.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.132739067077637, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649931.875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1455769538879395, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95067.8203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7287.962890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.973846435546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118617576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.604036961720399, + "step": 144, + "teaching_rms": 6.604074818469429, + "traffic_rms": 0.00018756098103314345 + }, + { + "batch_loss": 2.319704532623291, + "innovation_rms": 6.470993235599606, + "instruction_rms": 6.470933825062076, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 141361.609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 301761550090240.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97422.796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7287.26953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.005263805389404, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118607832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494173.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.419465065002441, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649932.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.431019306182861, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97422.796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7287.26220703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.005257606506348, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118607832.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.470933825062076, + "step": 145, + "teaching_rms": 6.470993235599606, + "traffic_rms": 0.0001867386459931176 + }, + { + "batch_loss": 2.312774896621704, + "innovation_rms": 6.8193961195707535, + "instruction_rms": 6.81936522997629, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 127225.4453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 271585395081216.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99541.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7286.56640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.033463954925537, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118597880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494173.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.777518272399902, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649933.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.787917137145996, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99541.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7286.55908203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.033458232879639, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118597880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.81936522997629, + "step": 146, + "teaching_rms": 6.8193961195707535, + "traffic_rms": 0.00018544303487903413 + }, + { + "batch_loss": 2.3247907161712646, + "innovation_rms": 6.861331635145091, + "instruction_rms": 6.861290553998377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 114502.8984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 244426857250816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101446.953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7285.8603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.058773994445801, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118587736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494173.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.199766159057617, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649933.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.209125518798828, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101446.953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7285.85302734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.058768272399902, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118587736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.861290553998377, + "step": 147, + "teaching_rms": 6.861331635145091, + "traffic_rms": 0.0001842958719622479 + }, + { + "batch_loss": 2.327329635620117, + "innovation_rms": 7.300601730704977, + "instruction_rms": 7.300575091903722, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 103052.609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 219984164814848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103161.03125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7285.15234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.081482410430908, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118577416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494174.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.6797895431518555, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649934.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.688212871551514, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103161.03125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7285.14501953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.08147668838501, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118577416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.300575114621989, + "step": 148, + "teaching_rms": 7.300601730704977, + "traffic_rms": 0.0001832040139532615 + }, + { + "batch_loss": 2.3139491081237793, + "innovation_rms": 6.841698894185613, + "instruction_rms": 6.841637526743438, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 92747.3515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 197985744977920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104702.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7284.4423828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.10184907913208, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118566944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494174.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.21181058883667, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649934.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.219391345977783, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104702.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7284.43505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.101843357086182, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118566944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.841637526743438, + "step": 149, + "teaching_rms": 6.841698894185613, + "traffic_rms": 0.00018268739452353494 + }, + { + "batch_loss": 2.2990522384643555, + "innovation_rms": 6.608962557780383, + "instruction_rms": 6.6089537739354, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 83472.6171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 178187170480128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 106089.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7283.73046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.120108127593994, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118556336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494174.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7906293869018555, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649934.875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.797452211380005, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 106089.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7283.72314453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.120102405548096, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118556336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.6089537739354, + "step": 150, + "teaching_rms": 6.608962557780383, + "traffic_rms": 0.0001824804022338585 + }, + { + "batch_loss": 2.303906202316284, + "innovation_rms": 7.157661279151881, + "instruction_rms": 7.1576560299280105, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 75125.3515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 160368458465280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107335.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7283.0166015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.136470317840576, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118545600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494174.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4115662574768066, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649935.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4177069664001465, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107335.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7283.00927734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1364641189575195, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118545600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.1576560299280105, + "step": 151, + "teaching_rms": 7.157661279151881, + "traffic_rms": 0.00018277427301021046 + }, + { + "batch_loss": 2.33156156539917, + "innovation_rms": 6.385262725809207, + "instruction_rms": 6.385195883488212, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 67612.8125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144331604230144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108456.7890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7282.30126953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.151124477386475, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118534752.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0704095363616943, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649935.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0759360790252686, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108456.7890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7282.2939453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.151118755340576, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118534752.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.385195896475786, + "step": 152, + "teaching_rms": 6.385262725809207, + "traffic_rms": 0.000183553776928683 + }, + { + "batch_loss": 2.3265092372894287, + "innovation_rms": 6.966021440485829, + "instruction_rms": 6.965970251775105, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 60851.53125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 129898442129408.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109464.5859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281.5849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.164241790771484, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118523808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.763368606567383, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649935.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7683424949645996, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109464.5859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281.57763671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.164236068725586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118523808.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.965970251775105, + "step": 153, + "teaching_rms": 6.966021440485829, + "traffic_rms": 0.000184033494766314 + }, + { + "batch_loss": 2.316887855529785, + "innovation_rms": 7.02741472405616, + "instruction_rms": 7.027374072127541, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 54766.37890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 116908598755328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110370.5078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7280.8671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.175975799560547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118512768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4870316982269287, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4915082454681396, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110370.5078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7280.85986328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.175970077514648, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118512768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.027374072127541, + "step": 154, + "teaching_rms": 7.02741472405616, + "traffic_rms": 0.00018446034899199818 + }, + { + "batch_loss": 2.315358877182007, + "innovation_rms": 6.928222924728833, + "instruction_rms": 6.928177620733657, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 49289.7421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 105217739718656.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111184.7265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7280.1484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.186464309692383, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118501648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.238328456878662, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649936.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2423572540283203, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111184.7265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7280.14111328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.186459064483643, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118501648.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.928177620733657, + "step": 155, + "teaching_rms": 6.928222924728833, + "traffic_rms": 0.00018498338007124206 + }, + { + "batch_loss": 2.355581045150757, + "innovation_rms": 6.93932036296734, + "instruction_rms": 6.939245087220877, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 44360.76953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 94695967424512.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111916.4140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7279.4287109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1958327293396, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118490456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.014495611190796, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649936.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0181214809417725, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111916.4140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7279.42138671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.195827007293701, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118490456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.939245087220877, + "step": 156, + "teaching_rms": 6.93932036296734, + "traffic_rms": 0.00018537027359534142 + }, + { + "batch_loss": 2.3116953372955322, + "innovation_rms": 6.416527987443822, + "instruction_rms": 6.41648977200439, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 39924.69140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 85226369843200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112573.8203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7278.70849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204192161560059, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118479200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8130459785461426, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649936.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8163093328475952, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112573.8203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7278.701171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204185962677002, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118479200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.41648977200439, + "step": 157, + "teaching_rms": 6.416527987443822, + "traffic_rms": 0.00018522083579911668 + }, + { + "batch_loss": 2.2972116470336914, + "innovation_rms": 6.820924501017679, + "instruction_rms": 6.820884477370743, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35932.22265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76703728664576.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113164.359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7277.9873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.211643695831299, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118467880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494175.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6317412853240967, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649936.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6346783638000488, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113164.359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7277.97998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.211637020111084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118467880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.820884489528726, + "step": 158, + "teaching_rms": 6.820924501017679, + "traffic_rms": 0.00018474701709287646 + }, + { + "batch_loss": 2.3251304626464844, + "innovation_rms": 6.637552130387107, + "instruction_rms": 6.637491799241756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32339.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69033357475840.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113694.7109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7277.265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.218277931213379, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118456512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.468567132949829, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649936.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4712104797363281, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113694.7109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7277.25830078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.218271255493164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118456512.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.637491799241756, + "step": 159, + "teaching_rms": 6.637552130387107, + "traffic_rms": 0.00018421343415213687 + }, + { + "batch_loss": 2.296638250350952, + "innovation_rms": 7.06949436765871, + "instruction_rms": 7.069449992253676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29105.099609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 62130019631104.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114170.8828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7276.54296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.224176406860352, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118445096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3217103481292725, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3240894079208374, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114170.8828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7276.53564453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.224169731140137, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118445096.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.069449992253676, + "step": 160, + "teaching_rms": 7.06949436765871, + "traffic_rms": 0.00018355705549608485 + }, + { + "batch_loss": 2.320828437805176, + "innovation_rms": 6.7909260471548025, + "instruction_rms": 6.790898913478453, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26194.58984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 55917018087424.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114598.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7275.81982421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.229413032531738, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118433640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1895393133163452, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.1875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1916804313659668, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114598.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7275.8125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.229405879974365, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118433640.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.790898913478453, + "step": 161, + "teaching_rms": 6.7909260471548025, + "traffic_rms": 0.00018319310008196987 + }, + { + "batch_loss": 2.3603129386901855, + "innovation_rms": 6.508164360507072, + "instruction_rms": 6.50817018726885, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 23575.130859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 50325314600960.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114981.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7275.09619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.234053611755371, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118422144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0705853700637817, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0725123882293701, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114981.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7275.0888671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23404598236084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118422144.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.50817018726885, + "step": 162, + "teaching_rms": 6.508164360507072, + "traffic_rms": 0.00018320309661951583 + }, + { + "batch_loss": 2.2941160202026367, + "innovation_rms": 7.122924963197873, + "instruction_rms": 7.1228787895062515, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21217.6171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 45292783140864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115325.8515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7274.37255859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.238157749176025, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118410608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9635267853736877, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9652611017227173, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115325.8515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7274.365234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.238150119781494, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118410608.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.1228787895062515, + "step": 163, + "teaching_rms": 7.122924963197873, + "traffic_rms": 0.00018336950404518272 + }, + { + "batch_loss": 2.339815378189087, + "innovation_rms": 6.389245239321946, + "instruction_rms": 6.389198645406389, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19095.85546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 40763505246208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115634.328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7273.6484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.241779327392578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118399048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8671740889549255, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8687349557876587, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115634.328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7273.64111328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.241771697998047, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118399048.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.389198645406389, + "step": 164, + "teaching_rms": 6.389245239321946, + "traffic_rms": 0.00018388125022755333 + }, + { + "batch_loss": 2.370469808578491, + "innovation_rms": 7.453937154647178, + "instruction_rms": 7.4539135332647355, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17186.26953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 36687153463296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115910.796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7272.923828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.24496603012085, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118387456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7804566621780396, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.781861424446106, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115910.796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7272.91650390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.244958400726318, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118387456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.4539135332647355, + "step": 165, + "teaching_rms": 7.453937154647178, + "traffic_rms": 0.00018444940896492632 + }, + { + "batch_loss": 2.2865521907806396, + "innovation_rms": 7.019336749680519, + "instruction_rms": 7.019341273138491, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15467.642578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 33018437697536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116158.453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7272.19921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2477617263793945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118375840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7024109959602356, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7036752700805664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116158.453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7272.19189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.247754096984863, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118375840.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.019341273138491, + "step": 166, + "teaching_rms": 7.019336749680519, + "traffic_rms": 0.0001851783058912137 + }, + { + "batch_loss": 2.329319953918457, + "innovation_rms": 7.172059364136785, + "instruction_rms": 7.172017506741768, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13920.8779296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 29716593508352.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116380.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7271.47412109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2502055168151855, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118364200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6321699023246765, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.6875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6333077549934387, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116380.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7271.466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.250197410583496, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118364200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.172017506741768, + "step": 167, + "teaching_rms": 7.172059364136785, + "traffic_rms": 0.00018610975346039518 + }, + { + "batch_loss": 2.382507562637329, + "innovation_rms": 6.415824294756265, + "instruction_rms": 6.415796970830924, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12528.7900390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 26744933318656.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116578.5859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7270.7490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2523322105407715, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118352544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5689529180526733, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5699769854545593, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116578.5859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7270.74169921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.252324104309082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118352544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.415796970830924, + "step": 168, + "teaching_rms": 6.415824294756265, + "traffic_rms": 0.00018666154384718988 + }, + { + "batch_loss": 2.3402702808380127, + "innovation_rms": 6.47386889313271, + "instruction_rms": 6.473850449014871, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11275.9111328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24070439567360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116755.984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7270.0234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.254173278808594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118340872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5147121548652649, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.8125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.51297926902771, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116755.984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7270.01611328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2541656494140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118340872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.473850449014871, + "step": 169, + "teaching_rms": 6.47386889313271, + "traffic_rms": 0.00018662634121647344 + }, + { + "batch_loss": 2.325784683227539, + "innovation_rms": 6.880389476983305, + "instruction_rms": 6.88031246237767, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10148.3203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 21663395610624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116914.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7269.2978515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.255757808685303, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118329184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4608518183231354, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4616813361644745, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116914.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7269.29052734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.25575065612793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118329184.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.880312474430641, + "step": 170, + "teaching_rms": 6.880389476983305, + "traffic_rms": 0.00018600827006168714 + }, + { + "batch_loss": 2.3201255798339844, + "innovation_rms": 6.413705142446999, + "instruction_rms": 6.413654370659298, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9133.48828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19497056468992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117055.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7268.572265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.257111549377441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118317480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4307659864425659, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649937.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41551318764686584, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117055.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7268.56494140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.257104396820068, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118317480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.413654383589244, + "step": 171, + "teaching_rms": 6.413705142446999, + "traffic_rms": 0.0001849203750289531 + }, + { + "batch_loss": 2.3079400062561035, + "innovation_rms": 6.415212766605503, + "instruction_rms": 6.415194373858286, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8220.1396484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17547351031808.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117182.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7267.8466796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2582573890686035, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118305760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46477848291397095, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.373961865901947, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117182.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7267.83935546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2582502365112305, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118305760.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.415194386785128, + "step": 172, + "teaching_rms": 6.415212766605503, + "traffic_rms": 0.0001836086364163881 + }, + { + "batch_loss": 2.3207814693450928, + "innovation_rms": 6.993409084683055, + "instruction_rms": 6.993386483910532, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7398.12548828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15792615718912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117294.4609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7267.12060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.25921630859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118294032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4287757873535156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33656567335128784, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117294.4609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7267.11328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259208679199219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118294032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.993386483910532, + "step": 173, + "teaching_rms": 6.993409084683055, + "traffic_rms": 0.00018260002128650985 + }, + { + "batch_loss": 2.312185049057007, + "innovation_rms": 6.939280531366984, + "instruction_rms": 6.939250738141625, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6658.31298828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14213354356736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117394.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7266.39453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260006904602051, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118282296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3327052891254425, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30290910601615906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117394.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7266.38720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259998798370361, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118282296.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.939250738141625, + "step": 174, + "teaching_rms": 6.939280531366984, + "traffic_rms": 0.00018215790675819082 + }, + { + "batch_loss": 2.322450876235962, + "innovation_rms": 6.216901417165734, + "instruction_rms": 6.2168342962658985, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5992.4814453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12792019025920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117483.203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7265.66845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260645389556885, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118270544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2721283733844757, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2726181745529175, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117483.203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7265.6611328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2606377601623535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118270544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.2168342962658985, + "step": 175, + "teaching_rms": 6.216901417165734, + "traffic_rms": 0.00018233919554836513 + }, + { + "batch_loss": 2.337726593017578, + "innovation_rms": 6.845895110326371, + "instruction_rms": 6.845838430766789, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5393.2333984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11512817123328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117561.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7264.9423828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261147499084473, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118258784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32048001885414124, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24535635113716125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117561.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7264.93505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261139869689941, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118258784.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.845838430766789, + "step": 176, + "teaching_rms": 6.845895110326371, + "traffic_rms": 0.0001825906279004436 + }, + { + "batch_loss": 2.387932777404785, + "innovation_rms": 6.92842923212916, + "instruction_rms": 6.928388968196657, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4853.91015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10361535201280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117631.6171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7264.21630859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261527061462402, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118247024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47467684745788574, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22082071006298065, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117631.6171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7264.208984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261518955230713, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118247024.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.928388968196657, + "step": 177, + "teaching_rms": 6.92842923212916, + "traffic_rms": 0.00018309259849949634 + }, + { + "batch_loss": 2.3621747493743896, + "innovation_rms": 6.944213696130634, + "instruction_rms": 6.94417932822675, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4368.51904296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9325381681152.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117693.15625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7263.490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261795997619629, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118235256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47550085186958313, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19873863458633423, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117693.15625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7263.48291015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2617878913879395, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118235256.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.94417932822675, + "step": 178, + "teaching_rms": 6.944213696130634, + "traffic_rms": 0.00018370466500286547 + }, + { + "batch_loss": 2.2875139713287354, + "innovation_rms": 6.654865959178554, + "instruction_rms": 6.654822608131091, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3931.667236328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8392843722752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117747.359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7262.76416015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261965274810791, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118223480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2655320167541504, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1788647621870041, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117747.359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7262.7568359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261957168579102, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118223480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.654822608131091, + "step": 179, + "teaching_rms": 6.654865959178554, + "traffic_rms": 0.00018434620044891308 + }, + { + "batch_loss": 2.3062243461608887, + "innovation_rms": 6.756569842816786, + "instruction_rms": 6.756523535941552, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3538.50048828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7553559298048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117794.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7262.0380859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.262044906616211, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118211704.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22237877547740936, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1609782874584198, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117794.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7262.03076171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.26203727722168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118211704.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.756523548215349, + "step": 180, + "teaching_rms": 6.756569842816786, + "traffic_rms": 0.00018483035349186068 + }, + { + "batch_loss": 2.315061330795288, + "innovation_rms": 6.804207334811508, + "instruction_rms": 6.804215415798512, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3184.650390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6798203420672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117836.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7261.31201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2620439529418945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118199920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5856512784957886, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14488045871257782, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117836.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7261.3046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.262036323547363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118199920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.804215427986281, + "step": 181, + "teaching_rms": 6.804207334811508, + "traffic_rms": 0.000185012249655047 + }, + { + "batch_loss": 2.298367738723755, + "innovation_rms": 7.1584458583171395, + "instruction_rms": 7.158396240957849, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2939.978271484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6118383288320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117872.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7260.5859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261970520019531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118188136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3394421637058258, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1303924024105072, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117872.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7260.57861328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261962890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118188136.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.158396240957849, + "step": 182, + "teaching_rms": 7.1584458583171395, + "traffic_rms": 0.00018523696044798866 + }, + { + "batch_loss": 2.3443071842193604, + "innovation_rms": 6.0872883202203285, + "instruction_rms": 6.0872109571251904, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2903.08349609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5506544959488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117904.484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7259.85986328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261831760406494, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118176344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36622849106788635, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.117353156208992, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117904.484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7259.8525390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261824607849121, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118176344.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.0872109571251904, + "step": 183, + "teaching_rms": 6.0872883202203285, + "traffic_rms": 0.00018543718410668876 + }, + { + "batch_loss": 2.346350908279419, + "innovation_rms": 6.122509062040802, + "instruction_rms": 6.122434378983372, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2828.35546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4955890515968.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117931.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7259.1337890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261634349822998, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118164552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33376917243003845, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10561783611774445, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117931.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7259.12646484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261627197265625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118164552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.122434378983372, + "step": 184, + "teaching_rms": 6.122509062040802, + "traffic_rms": 0.00018509929203591665 + }, + { + "batch_loss": 2.330603837966919, + "innovation_rms": 6.968899695219279, + "instruction_rms": 6.968886426406697, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2844.864501953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4460301516800.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117954.9453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7258.40771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2613844871521, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118152760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3796825110912323, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09505604952573776, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117954.9453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7258.400390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261376857757568, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118152760.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.968886426406697, + "step": 185, + "teaching_rms": 6.968899695219279, + "traffic_rms": 0.00018429715263629853 + }, + { + "batch_loss": 2.305657386779785, + "innovation_rms": 6.660024655119301, + "instruction_rms": 6.65999353855846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2918.9794921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4014271365120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117974.71875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7257.681640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.261086463928223, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118140960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3711543083190918, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08555044233798981, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117974.71875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7257.67431640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.26107931137085, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118140960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.65999353855846, + "step": 186, + "teaching_rms": 6.660024655119301, + "traffic_rms": 0.0001836904980226516 + }, + { + "batch_loss": 2.3280744552612305, + "innovation_rms": 6.7220673525866115, + "instruction_rms": 6.72204900729865, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2874.620849609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3612844228608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117991.3359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7256.95556640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260745525360107, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118129160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33078837394714355, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07699539512395859, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117991.3359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7256.9482421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260738849639893, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118129160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.72204900729865, + "step": 187, + "teaching_rms": 6.7220673525866115, + "traffic_rms": 0.00018336954447915896 + }, + { + "batch_loss": 2.3350839614868164, + "innovation_rms": 6.864524098054408, + "instruction_rms": 6.864526912104889, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2810.32177734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3251559727104.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118005.1171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7256.2294921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260366439819336, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118117360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30717262625694275, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06929585337638855, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118005.1171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7256.22216796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260359287261963, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118117360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.864526912104889, + "step": 188, + "teaching_rms": 6.864524098054408, + "traffic_rms": 0.00018335141568567322 + }, + { + "batch_loss": 2.318922519683838, + "innovation_rms": 6.664936532166203, + "instruction_rms": 6.664926865688292, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2805.90234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2926403649536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118016.3359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7255.50341796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259952545166016, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118105560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5080845355987549, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.062366265803575516, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118016.3359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7255.49609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259945392608643, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118105560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.664926865688292, + "step": 189, + "teaching_rms": 6.664936532166203, + "traffic_rms": 0.00018372379474670616 + }, + { + "batch_loss": 2.29748272895813, + "innovation_rms": 6.747587248554058, + "instruction_rms": 6.747542230101712, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2846.233154296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2633763389440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118025.2578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7254.77734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259507179260254, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118093760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3647289276123047, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.056129638105630875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118025.2578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7254.77001953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259500503540039, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118093760.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.747542230101712, + "step": 190, + "teaching_rms": 6.747587248554058, + "traffic_rms": 0.00018425500641171234 + }, + { + "batch_loss": 2.3458099365234375, + "innovation_rms": 6.666206322684193, + "instruction_rms": 6.666132506829984, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2890.885498046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2370387050496.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118032.1015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7254.05126953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259034156799316, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118081960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29276955127716064, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05051667243242264, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118032.1015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7254.0439453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.259027481079102, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118081960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.666132506829984, + "step": 191, + "teaching_rms": 6.666206322684193, + "traffic_rms": 0.00018492027479174306 + }, + { + "batch_loss": 2.374786615371704, + "innovation_rms": 6.923849856119739, + "instruction_rms": 6.923757501474037, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2841.950439453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2133348319232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118037.0859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7253.3251953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.258535861968994, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118070160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3321911096572876, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.045465003699064255, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118037.0859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7253.31787109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.258528709411621, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118070160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.923757501474037, + "step": 192, + "teaching_rms": 6.923849856119739, + "traffic_rms": 0.00018511429911956546 + }, + { + "batch_loss": 2.337264060974121, + "innovation_rms": 6.473863959953946, + "instruction_rms": 6.473832987222155, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2706.416015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1920013434880.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118040.390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7252.599609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.258014678955078, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118058352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4984857738018036, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0409185029566288, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118040.390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7252.59228515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.258007526397705, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118058352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.473832987222155, + "step": 193, + "teaching_rms": 6.473863959953946, + "traffic_rms": 0.0001848394258096442 + }, + { + "batch_loss": 2.3261525630950928, + "innovation_rms": 6.585591866592894, + "instruction_rms": 6.585538827373063, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2649.073974609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1728012091392.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118042.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7251.8740234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.257472991943359, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118046544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4757539629936218, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03682665154337883, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118042.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7251.86669921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.257465839385986, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118046544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.585538827373063, + "step": 194, + "teaching_rms": 6.585591866592894, + "traffic_rms": 0.0001846671763056061 + }, + { + "batch_loss": 2.3077523708343506, + "innovation_rms": 7.095236335106038, + "instruction_rms": 7.095212139843689, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2733.48095703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1555210829824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118042.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7251.1484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.256912708282471, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118034736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3041595220565796, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03314398601651192, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118042.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7251.14111328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.256906032562256, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118034736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.095212139843689, + "step": 195, + "teaching_rms": 7.095236335106038, + "traffic_rms": 0.00018456662709658733 + }, + { + "batch_loss": 2.3637051582336426, + "innovation_rms": 6.590672513371866, + "instruction_rms": 6.590647235889782, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2723.800537109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1399689707520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118041.8359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7250.4228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.256336212158203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118022928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32162055373191833, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.029829585924744606, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118041.8359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7250.41552734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.256329536437988, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118022928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.590647235889782, + "step": 196, + "teaching_rms": 6.590672513371866, + "traffic_rms": 0.00018466381037937523 + }, + { + "batch_loss": 2.317683458328247, + "innovation_rms": 6.1233952609520825, + "instruction_rms": 6.123348338678624, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2720.869384765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1259720671232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118039.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7249.697265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.255744934082031, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118011128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44619643688201904, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0268466267734766, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118039.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7249.68994140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.255737781524658, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118011128.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.123348352221574, + "step": 197, + "teaching_rms": 6.1233952609520825, + "traffic_rms": 0.00018444035766207952 + }, + { + "batch_loss": 2.2983810901641846, + "innovation_rms": 6.821059076194965, + "instruction_rms": 6.821030846545377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2665.6337890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1133748551680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118037.046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7248.9716796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.25514030456543, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117999328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5042240619659424, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.024161962792277336, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118037.046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7248.96435546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.255132675170898, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117999328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.821030846545377, + "step": 198, + "teaching_rms": 6.821059076194965, + "traffic_rms": 0.00018391581590349685 + }, + { + "batch_loss": 2.3234493732452393, + "innovation_rms": 6.629796884981649, + "instruction_rms": 6.629771645951061, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2739.925048828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1020373696512.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118033.265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7248.24609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.254523277282715, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117987528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.48794588446617126, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.021745765581727028, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118033.265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7248.23876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.254515647888184, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117987528.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.629771645951061, + "step": 199, + "teaching_rms": 6.629796884981649, + "traffic_rms": 0.0001836001292213626 + }, + { + "batch_loss": 2.3373961448669434, + "innovation_rms": 7.172954234351531, + "instruction_rms": 7.172917470503258, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2678.84814453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 918336307200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118028.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7247.5205078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.253895282745361, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117975728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31953567266464233, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01957118883728981, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118028.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7247.51318359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.25388765335083, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117975728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.172917470503258, + "step": 200, + "teaching_rms": 7.172954234351531, + "traffic_rms": 0.00018339647152728448 + }, + { + "batch_loss": 2.3462865352630615, + "innovation_rms": 6.205236707250185, + "instruction_rms": 6.205174607000871, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2749.057861328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 826502676480.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118023.3828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7246.79541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.253257751464844, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117963928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26081860065460205, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01761407032608986, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118023.3828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7246.7880859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2532501220703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117963928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.205174607000871, + "step": 201, + "teaching_rms": 6.205236707250185, + "traffic_rms": 0.00018343155970511332 + }, + { + "batch_loss": 2.3425025939941406, + "innovation_rms": 6.940744129258521, + "instruction_rms": 6.9406862289064355, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2796.465087890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 743852408832.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118017.4296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7246.0703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.25261116027832, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117952128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37534722685813904, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.015852663666009903, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118017.4296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7246.06298828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.252604007720947, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117952128.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.9406862289064355, + "step": 202, + "teaching_rms": 6.940744129258521, + "traffic_rms": 0.00018328865869596752 + }, + { + "batch_loss": 2.3094093799591064, + "innovation_rms": 6.8640887908500785, + "instruction_rms": 6.864062827794249, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2712.92431640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 669467148288.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118010.8984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7245.34521484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.251956939697266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117940328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4552956223487854, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.014267397113144398, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118010.8984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7245.337890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.251949310302734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117940328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.864062839875753, + "step": 203, + "teaching_rms": 6.8640887908500785, + "traffic_rms": 0.00018322930627307995 + }, + { + "batch_loss": 2.3195858001708984, + "innovation_rms": 6.7295246424482436, + "instruction_rms": 6.729518085896842, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2705.95068359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 602520420352.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118003.8359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7244.6201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.251295566558838, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117928528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2502126693725586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.012840657494962215, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118003.8359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7244.61279296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.251288414001465, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117928528.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.729518085896842, + "step": 204, + "teaching_rms": 6.7295246424482436, + "traffic_rms": 0.00018372919564642947 + }, + { + "batch_loss": 2.306997299194336, + "innovation_rms": 6.311649800283879, + "instruction_rms": 6.311586275529492, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2725.937744140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 542268391424.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117996.3046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7243.89501953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2506279945373535, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117916728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11394746601581573, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011556591838598251, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117996.3046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7243.8876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.250620365142822, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117916728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.311586275529492, + "step": 205, + "teaching_rms": 6.311649800283879, + "traffic_rms": 0.00018451246767087494 + }, + { + "batch_loss": 2.3084263801574707, + "innovation_rms": 6.443836851210974, + "instruction_rms": 6.4437998771637695, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2762.43212890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 488041545728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117988.34375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7243.169921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.249954700469971, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117904928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33907416462898254, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.010400932282209396, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117988.34375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7243.16259765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.249946594238281, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117904928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.443799902902682, + "step": 206, + "teaching_rms": 6.443836851210974, + "traffic_rms": 0.0001849143272866901 + }, + { + "batch_loss": 2.317605972290039, + "innovation_rms": 7.059294171103857, + "instruction_rms": 7.0592831510920195, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2739.083740234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 439237378048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117980.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7242.44482421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.249276161193848, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117893128.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39499014616012573, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.009360838681459427, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117980.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7242.4375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.249268054962158, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117893128.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.0592831510920195, + "step": 207, + "teaching_rms": 7.059294171103857, + "traffic_rms": 0.0001851691767682845 + }, + { + "batch_loss": 2.292736530303955, + "innovation_rms": 6.876109404656992, + "instruction_rms": 6.876062972351626, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2691.07861328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 395313643520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117971.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7241.7197265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.248592853546143, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117881328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4723494052886963, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00842475425451994, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117971.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7241.71240234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.248584747314453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117881328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.876062972351626, + "step": 208, + "teaching_rms": 6.876109404656992, + "traffic_rms": 0.00018555811363456226 + }, + { + "batch_loss": 2.294790506362915, + "innovation_rms": 6.789851850941021, + "instruction_rms": 6.789813464607469, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2742.525390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 355782262784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117962.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7240.9951171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.247905731201172, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117869528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30257007479667664, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.007582278456538916, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117962.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7240.98779296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.247897148132324, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117869528.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.789813464607469, + "step": 209, + "teaching_rms": 6.789851850941021, + "traffic_rms": 0.00018579250694958336 + }, + { + "batch_loss": 2.312810182571411, + "innovation_rms": 6.599597716415624, + "instruction_rms": 6.599586120103706, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2705.63134765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 320204046336.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117953.03125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7240.2705078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2472147941589355, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117857736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18170930445194244, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.006824050564318895, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117953.03125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7240.26318359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.24720573425293, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117857736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.599586120103706, + "step": 210, + "teaching_rms": 6.599597716415624, + "traffic_rms": 0.00018575917524488072 + }, + { + "batch_loss": 2.3206348419189453, + "innovation_rms": 6.484438267760016, + "instruction_rms": 6.484368840511182, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2813.971923828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 288183648256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117943.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7239.5458984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.246520519256592, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117845944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3161879777908325, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00614164536818862, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117943.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7239.53857421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.246510982513428, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117845944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.484368840511182, + "step": 211, + "teaching_rms": 6.484438267760016, + "traffic_rms": 0.0001853868651747304 + }, + { + "batch_loss": 2.3193728923797607, + "innovation_rms": 6.645564123380986, + "instruction_rms": 6.645503153416238, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2780.591796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 259365289984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117933.7421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7238.8212890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.245822906494141, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117834152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4198176860809326, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0055274805054068565, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117933.7421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7238.81396484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.245813369750977, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117834152.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.645503153416238, + "step": 212, + "teaching_rms": 6.645564123380986, + "traffic_rms": 0.0001845268595041916 + }, + { + "batch_loss": 2.319516181945801, + "innovation_rms": 6.3498771631435105, + "instruction_rms": 6.349850429778116, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2693.371826171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 233428762624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117923.78125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7238.0966796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.24512243270874, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117822360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31893524527549744, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0049747321754693985, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117923.78125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7238.08935546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.245112895965576, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117822360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.349850429778116, + "step": 213, + "teaching_rms": 6.3498771631435105, + "traffic_rms": 0.0001836677316974276 + }, + { + "batch_loss": 2.328481674194336, + "innovation_rms": 6.5290215420912965, + "instruction_rms": 6.528990639496344, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2594.984130859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 210085888000.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117913.6328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7237.3720703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.244419574737549, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117810568.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2257899045944214, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004477258771657944, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117913.6328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7237.36474609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.244410037994385, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117810568.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.528990639496344, + "step": 214, + "teaching_rms": 6.5290215420912965, + "traffic_rms": 0.00018320857357342953 + }, + { + "batch_loss": 2.307856798171997, + "innovation_rms": 6.709297173185594, + "instruction_rms": 6.7092704625122375, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2589.169677734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 189077291008.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117903.3203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7236.6474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.243714809417725, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117798776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2023727148771286, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004029532894492149, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117903.3203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7236.64013671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2437052726745605, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117798776.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.7092704625122375, + "step": 215, + "teaching_rms": 6.709297173185594, + "traffic_rms": 0.00018311557804496636 + }, + { + "batch_loss": 2.311102867126465, + "innovation_rms": 6.3056307026533345, + "instruction_rms": 6.305597601550568, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2534.11572265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 170169565184.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117892.859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7235.92333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.243008136749268, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117786984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2394673377275467, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0036265794187784195, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117892.859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7235.916015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2429986000061035, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117786984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.305597601550568, + "step": 216, + "teaching_rms": 6.3056307026533345, + "traffic_rms": 0.00018339741484567666 + }, + { + "batch_loss": 2.3280556201934814, + "innovation_rms": 6.533689442931576, + "instruction_rms": 6.533621793575537, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2561.117919921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 153152602112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117882.265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7235.19921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.242299556732178, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117775192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36320960521698, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0032639214769005775, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117882.265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7235.19189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.242290496826172, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117775192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.533621793575537, + "step": 217, + "teaching_rms": 6.533689442931576, + "traffic_rms": 0.0001837784286586776 + }, + { + "batch_loss": 2.3137314319610596, + "innovation_rms": 5.931965552932883, + "instruction_rms": 5.931912320535702, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2589.4091796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 137837346816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117871.5546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7234.47509765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.241589069366455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117763408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39719754457473755, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0029375292360782623, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117871.5546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7234.4677734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.241580486297607, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117763408.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.931912320535702, + "step": 218, + "teaching_rms": 5.931965552932883, + "traffic_rms": 0.0001840021210951526 + }, + { + "batch_loss": 2.31494140625, + "innovation_rms": 6.841479004512189, + "instruction_rms": 6.841449379857457, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2688.449462890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 124053610496.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117860.7421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7233.7509765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.240877151489258, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117751624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35452038049697876, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002643776126205921, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117860.7421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7233.74365234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.240869045257568, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117751624.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.841449379857457, + "step": 219, + "teaching_rms": 6.841479004512189, + "traffic_rms": 0.00018407810171118866 + }, + { + "batch_loss": 2.3405303955078125, + "innovation_rms": 6.293111531133399, + "instruction_rms": 6.293088670281488, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2575.49609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 111648251904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117849.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7233.02685546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.240164756774902, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117739840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16905219852924347, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002379398560151458, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117849.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7233.01953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.240156173706055, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117739840.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.293088670281488, + "step": 220, + "teaching_rms": 6.293111531133399, + "traffic_rms": 0.000184384626419789 + }, + { + "batch_loss": 2.278815269470215, + "innovation_rms": 6.281480594126556, + "instruction_rms": 6.281431865751881, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2571.996337890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 100483424256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117838.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7232.302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.239450931549072, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117728056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12625938653945923, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0021414586808532476, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117838.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7232.29541015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.239442348480225, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117728056.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.281431865751881, + "step": 221, + "teaching_rms": 6.281480594126556, + "traffic_rms": 0.0001845945242482466 + }, + { + "batch_loss": 2.323031187057495, + "innovation_rms": 6.286699293206527, + "instruction_rms": 6.2866266521289145, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2518.946044921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 90435084288.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117827.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7231.5791015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.238736152648926, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117716272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2856203317642212, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0019273128127679229, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117827.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7231.57177734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.238727569580078, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117716272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.2866266521289145, + "step": 222, + "teaching_rms": 6.286699293206527, + "traffic_rms": 0.000184712714001399 + }, + { + "batch_loss": 2.326785087585449, + "innovation_rms": 6.6410031302170545, + "instruction_rms": 6.640991566936926, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2508.82177734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81391575040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117816.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7230.85546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.238020420074463, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117704488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30883872509002686, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0017345815431326628, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117816.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7230.84814453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.238011360168457, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117704488.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.640991566936926, + "step": 223, + "teaching_rms": 6.6410031302170545, + "traffic_rms": 0.00018451288959726983 + }, + { + "batch_loss": 2.306290626525879, + "innovation_rms": 6.705812549749075, + "instruction_rms": 6.705756615394034, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2515.1015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73252413440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117805.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7230.1318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.237303733825684, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117692704.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4544892907142639, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.001561123295687139, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117805.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7230.12451171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.237295150756836, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117692704.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.705756615394034, + "step": 224, + "teaching_rms": 6.705812549749075, + "traffic_rms": 0.00018446643171473847 + }, + { + "batch_loss": 2.3016061782836914, + "innovation_rms": 5.88429101474866, + "instruction_rms": 5.884254163354303, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2592.52685546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 65927172096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117794.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7229.408203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.236586570739746, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117680928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33579695224761963, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0014050109311938286, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117794.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7229.40087890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.236577987670898, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117680928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.884254163354303, + "step": 225, + "teaching_rms": 5.88429101474866, + "traffic_rms": 0.00018460360815582475 + }, + { + "batch_loss": 2.339987277984619, + "innovation_rms": 7.034706263550451, + "instruction_rms": 7.034724036264097, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2577.614501953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 59334455296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117782.7890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7228.6845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.235868453979492, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117669152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22688287496566772, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00126450986135751, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117782.7890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7228.67724609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2358598709106445, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117669152.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.034724036264097, + "step": 226, + "teaching_rms": 7.034706263550451, + "traffic_rms": 0.00018459198710061008 + }, + { + "batch_loss": 2.3301796913146973, + "innovation_rms": 6.515890202909416, + "instruction_rms": 6.515851551704165, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2657.809814453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 53401010176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117771.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7227.9609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23514986038208, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117657376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34826767444610596, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0011380588402971625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117771.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7227.95361328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.235141754150391, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117657376.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.515851551704165, + "step": 227, + "teaching_rms": 6.515890202909416, + "traffic_rms": 0.0001850274641569236 + }, + { + "batch_loss": 2.3217670917510986, + "innovation_rms": 6.142840784799293, + "instruction_rms": 6.142800274846185, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2728.700927734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 48060907520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117760.0078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7227.2373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23443078994751, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117645600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36414635181427, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0010242529679089785, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117760.0078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7227.22998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23442268371582, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117645600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.142800274846185, + "step": 228, + "teaching_rms": 6.142840784799293, + "traffic_rms": 0.00018513920365402994 + }, + { + "batch_loss": 2.3554773330688477, + "innovation_rms": 5.878408366117118, + "instruction_rms": 5.878355749972768, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2730.6611328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 43254816768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117748.5703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7226.51416015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.233711242675781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117633824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22085221111774445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0009218276245519519, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117748.5703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7226.5068359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.233703136444092, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117633824.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.878355749972768, + "step": 229, + "teaching_rms": 5.878408366117118, + "traffic_rms": 0.0001848694007370171 + }, + { + "batch_loss": 2.3077495098114014, + "innovation_rms": 6.61886342987577, + "instruction_rms": 6.618814604481754, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2743.88134765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38929334272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117737.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7225.791015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2329912185668945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117622048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2509346008300781, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0008296448504552245, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117737.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7225.78369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.232983112335205, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117622048.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.618814604481754, + "step": 230, + "teaching_rms": 6.61886342987577, + "traffic_rms": 0.0001842581517850277 + }, + { + "batch_loss": 2.322310447692871, + "innovation_rms": 6.533493051778612, + "instruction_rms": 6.533432609219039, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2755.914794921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 35036401664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117725.5859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7225.06787109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23227071762085, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117610272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3105246424674988, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007466803654097021, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117725.5859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7225.060546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.232263088226318, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117610272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.533432609219039, + "step": 231, + "teaching_rms": 6.533493051778612, + "traffic_rms": 0.00018369307453006074 + }, + { + "batch_loss": 2.309081792831421, + "innovation_rms": 6.4399178697758, + "instruction_rms": 6.439867573294263, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2827.60986328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 31532761088.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117714.046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7224.3447265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.231550216674805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117598504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2646254301071167, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000672012334689498, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117714.046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7224.33740234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.231542587280273, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117598504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.439867573294263, + "step": 232, + "teaching_rms": 6.4399178697758, + "traffic_rms": 0.00018329418028559694 + }, + { + "batch_loss": 2.318525552749634, + "innovation_rms": 6.760904144963893, + "instruction_rms": 6.760908399873411, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2876.6806640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 28379484160.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117702.4921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7223.62158203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23082971572876, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117586736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15500348806381226, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0006048110662959516, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117702.4921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7223.6142578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.23082160949707, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117586736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.760908399873411, + "step": 233, + "teaching_rms": 6.760904144963893, + "traffic_rms": 0.00018303098828838246 + }, + { + "batch_loss": 2.3168275356292725, + "innovation_rms": 6.7292046791482525, + "instruction_rms": 6.729161521959976, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2842.26025390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 25541535744.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117690.9140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7222.8984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.230108737945557, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117574968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.256197452545166, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000544329930562526, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117690.9140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7222.89111328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.230100631713867, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117574968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.729161534283681, + "step": 234, + "teaching_rms": 6.7292046791482525, + "traffic_rms": 0.00018328362237576053 + }, + { + "batch_loss": 2.3287227153778076, + "innovation_rms": 6.247067250175316, + "instruction_rms": 6.2470350075985746, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2744.501220703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22987382784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117679.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7222.17529296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.229387283325195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117563200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2997933626174927, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0004898969200439751, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117679.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7222.16796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.229379653930664, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117563200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.2470350075985746, + "step": 235, + "teaching_rms": 6.247067250175316, + "traffic_rms": 0.0001837492441732997 + }, + { + "batch_loss": 2.308645009994507, + "innovation_rms": 6.196441997134758, + "instruction_rms": 6.196384479043028, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2735.51708984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20688644096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117667.6953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7221.45263671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.228665828704834, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117551432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20902198553085327, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0004409072280395776, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117667.6953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7221.4453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.228658199310303, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117551432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.196384492426348, + "step": 236, + "teaching_rms": 6.196441997134758, + "traffic_rms": 0.0001842295262379409 + }, + { + "batch_loss": 2.3069112300872803, + "innovation_rms": 6.1376472824732184, + "instruction_rms": 6.137566392695007, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2755.68408203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 18619779072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117656.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7220.72998046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.227944374084473, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117539664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1330595761537552, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003968164965044707, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117656.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7220.72265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.227936744689941, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117539664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.137566392695007, + "step": 237, + "teaching_rms": 6.1376472824732184, + "traffic_rms": 0.00018448436386109545 + }, + { + "batch_loss": 2.2766213417053223, + "innovation_rms": 6.900341177080988, + "instruction_rms": 6.900326441570825, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2784.502197265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 16757800960.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117644.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7220.00732421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.227222919464111, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117527896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09916031360626221, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003571348497644067, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117644.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7220.0, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.227214813232422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117527896.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.9003264535888365, + "step": 238, + "teaching_rms": 6.900341177080988, + "traffic_rms": 0.0001843313386504402 + }, + { + "batch_loss": 2.3357088565826416, + "innovation_rms": 6.688564649820762, + "instruction_rms": 6.688502595562416, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2795.29345703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15082020864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117632.765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7219.28466796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.226500988006592, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117516136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20206661522388458, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000321421364787966, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117632.765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7219.27734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2264933586120605, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117516136.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.688502595562416, + "step": 239, + "teaching_rms": 6.688564649820762, + "traffic_rms": 0.00018447690183488502 + }, + { + "batch_loss": 2.307896614074707, + "innovation_rms": 6.858145608794735, + "instruction_rms": 6.858115994889053, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2845.046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13573818368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117621.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7218.56201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2257795333862305, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117504376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3033469319343567, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00028927921084687114, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117621.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7218.5546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.225771427154541, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117504376.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.858115994889053, + "step": 240, + "teaching_rms": 6.858145608794735, + "traffic_rms": 0.00018446783848328954 + }, + { + "batch_loss": 2.2994472980499268, + "innovation_rms": 6.430860813831955, + "instruction_rms": 6.430800169363558, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2784.502197265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12216436736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117609.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7217.83935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.225057601928711, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117492616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26637864112854004, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00026035128394141793, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117609.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7217.83203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.22504997253418, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117492616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.430800169363558, + "step": 241, + "teaching_rms": 6.430860813831955, + "traffic_rms": 0.00018461048270987778 + }, + { + "batch_loss": 2.3057610988616943, + "innovation_rms": 6.748828367830764, + "instruction_rms": 6.748811778908734, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2814.651123046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10994793472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117597.734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7217.11669921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.224335670471191, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117480856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2293611615896225, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000234316146816127, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117597.734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7217.109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.22432804107666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117480856.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.748811778908734, + "step": 242, + "teaching_rms": 6.748828367830764, + "traffic_rms": 0.0001844377714939653 + }, + { + "batch_loss": 2.312324285507202, + "innovation_rms": 6.647552622624515, + "instruction_rms": 6.647536768610025, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2890.61474609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9895314432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117586.046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7216.39453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.223613739013672, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117469096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26615583896636963, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00021088452194817364, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117586.046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7216.38720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.223606109619141, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117469096.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.647536768610025, + "step": 243, + "teaching_rms": 6.647552622624515, + "traffic_rms": 0.00018440364541831046 + }, + { + "batch_loss": 2.320770740509033, + "innovation_rms": 6.579197918479375, + "instruction_rms": 6.579150375365365, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2843.51318359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8905783296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117574.34375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7215.67236328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.222891807556152, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117457336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.343375563621521, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00018979606102220714, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117574.34375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7215.6650390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.222884178161621, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117457336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.579150375365365, + "step": 244, + "teaching_rms": 6.579197918479375, + "traffic_rms": 0.00018434520165562962 + }, + { + "batch_loss": 2.3078579902648926, + "innovation_rms": 6.726584239544707, + "instruction_rms": 6.726525125357202, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2914.117431640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8015204864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117562.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7214.9501953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.222169399261475, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117445576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31434595584869385, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00017081644909922034, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117562.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7214.94287109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.222162246704102, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117445576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.726525125357202, + "step": 245, + "teaching_rms": 6.726584239544707, + "traffic_rms": 0.00018427433384850077 + }, + { + "batch_loss": 2.3164799213409424, + "innovation_rms": 6.494619015353291, + "instruction_rms": 6.494610510744442, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2858.816650390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7213684224.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117550.9296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7214.22802734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.221447467803955, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117433824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2424509972333908, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00015373479982372373, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117550.9296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7214.220703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.221439838409424, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117433824.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.49461049797567, + "step": 246, + "teaching_rms": 6.494619015353291, + "traffic_rms": 0.00018425241093530226 + }, + { + "batch_loss": 2.315408229827881, + "innovation_rms": 6.245787143233792, + "instruction_rms": 6.2457506574889115, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2767.302001953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6492315648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117539.21875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7213.505859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.220725059509277, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117422072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1437595635652542, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00013836131256539375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117539.21875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7213.49853515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.220717906951904, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117422072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.2457506574889115, + "step": 247, + "teaching_rms": 6.245787143233792, + "traffic_rms": 0.00018447210561560286 + }, + { + "batch_loss": 2.314312219619751, + "innovation_rms": 6.2621369366422615, + "instruction_rms": 6.262090996399853, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2761.845458984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5843084288.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117527.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7212.78369140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.220003128051758, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117410320.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15292944014072418, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0001245251769432798, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117527.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7212.7763671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.219995498657227, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117410320.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.262090996399853, + "step": 248, + "teaching_rms": 6.2621369366422615, + "traffic_rms": 0.00018466419180948566 + }, + { + "batch_loss": 2.3249454498291016, + "innovation_rms": 6.479541046079885, + "instruction_rms": 6.479507117555888, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2817.298828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5258776064.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117515.78125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7212.0615234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.21928071975708, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117398568.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.296479195356369, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00011207265924895182, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117515.78125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7212.05419921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.219273567199707, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117398568.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.479507117555888, + "step": 249, + "teaching_rms": 6.479541046079885, + "traffic_rms": 0.0001848740128612717 + }, + { + "batch_loss": 2.3020200729370117, + "innovation_rms": 6.62188901029244, + "instruction_rms": 6.6218472324018345, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2808.2890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4732898304.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117504.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7211.33984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2185587882995605, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117386816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3738989233970642, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00010086539259646088, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117504.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7211.33251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.218551158905029, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117386816.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.6218472324018345, + "step": 250, + "teaching_rms": 6.62188901029244, + "traffic_rms": 0.00018488797553311575 + }, + { + "batch_loss": 2.346102476119995, + "innovation_rms": 7.182598156479782, + "instruction_rms": 7.182575573886158, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2859.23095703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4259608576.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117492.3359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7210.6181640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.217836380004883, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117375064.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22309842705726624, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.077884897124022e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117492.3359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7210.61083984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.21782922744751, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117375064.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.1825755854319056, + "step": 251, + "teaching_rms": 7.182598156479782, + "traffic_rms": 0.00018486412004766798 + }, + { + "batch_loss": 2.3220279216766357, + "innovation_rms": 6.770958473952841, + "instruction_rms": 6.770931910049025, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2894.755859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3833647616.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117480.609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7209.896484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.217114448547363, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117363312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.214574933052063, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.17009640741162e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117480.609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7209.88916015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.21710729598999, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117363312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.770931922296703, + "step": 252, + "teaching_rms": 6.770958473952841, + "traffic_rms": 0.00018479214057439173 + }, + { + "batch_loss": 2.3091213703155518, + "innovation_rms": 6.620617966462752, + "instruction_rms": 6.620573652013762, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2839.15966796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3450282752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117468.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7209.1748046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.216392517089844, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117351568.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2440706193447113, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.353086402872577e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117468.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7209.16748046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.216385364532471, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117351568.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.620573652013762, + "step": 253, + "teaching_rms": 6.620617966462752, + "traffic_rms": 0.00018469041491353522 + }, + { + "batch_loss": 2.320317506790161, + "innovation_rms": 6.6207642277457985, + "instruction_rms": 6.620715443429827, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2811.588134765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3105254400.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117457.140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7208.453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.215670585632324, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117339824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4309018850326538, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.617777398787439e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117457.140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7208.44580078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.215663433074951, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117339824.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.620715443429827, + "step": 254, + "teaching_rms": 6.6207642277457985, + "traffic_rms": 0.00018448775985387486 + }, + { + "batch_loss": 2.3193581104278564, + "innovation_rms": 6.680657129230949, + "instruction_rms": 6.680613573105198, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2713.42333984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2794728960.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117445.40625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7207.7314453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.214948654174805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117328080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3438892960548401, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.955999586149119e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117445.40625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7207.72412109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.214941501617432, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117328080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.680613573105198, + "step": 255, + "teaching_rms": 6.680657129230949, + "traffic_rms": 0.00018419199374119117 + }, + { + "batch_loss": 2.312452554702759, + "innovation_rms": 7.175004099294974, + "instruction_rms": 7.174979767443306, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2717.962646484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2515256064.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117433.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7207.01025390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.214226722717285, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117316336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10968003422021866, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.360399518394843e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117433.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7207.0029296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.214219570159912, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117316336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.174979767443306, + "step": 256, + "teaching_rms": 7.175004099294974, + "traffic_rms": 0.00018382745354131555 + }, + { + "batch_loss": 2.344525098800659, + "innovation_rms": 6.498895238996828, + "instruction_rms": 6.498828326363946, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2665.90625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2263730432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117421.9296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7206.2890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.213504791259766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117304592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22959209978580475, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.82435934827663e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117421.9296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7206.28173828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.213497638702393, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117304592.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.498828326363946, + "step": 257, + "teaching_rms": 6.498895238996828, + "traffic_rms": 0.00018384705445672735 + }, + { + "batch_loss": 2.3125298023223877, + "innovation_rms": 6.470311280367587, + "instruction_rms": 6.470278637418909, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2715.404052734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2037357440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117410.1953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7205.56787109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.212782859802246, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117292848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.289000004529953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.341923340689391e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117410.1953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7205.560546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.212776184082031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117292848.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.470278637418909, + "step": 258, + "teaching_rms": 6.470311280367587, + "traffic_rms": 0.00018352259058360295 + }, + { + "batch_loss": 2.3199024200439453, + "innovation_rms": 5.94101369001869, + "instruction_rms": 5.94095636514713, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2638.995361328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833621632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117398.453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7204.8466796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.212060928344727, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117281112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24490416049957275, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.907730933860876e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117398.453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7204.83935546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.212054252624512, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117281112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.94095636514713, + "step": 259, + "teaching_rms": 5.94101369001869, + "traffic_rms": 0.0001833193253485195 + }, + { + "batch_loss": 2.3164913654327393, + "innovation_rms": 6.773040078105713, + "instruction_rms": 6.773018371461168, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2723.488037109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1650259456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117386.7109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7204.12548828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.211339473724365, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117269376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28705790638923645, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.516957804095e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117386.7109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7204.1181640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.211332321166992, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117269376.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.773018371461168, + "step": 260, + "teaching_rms": 6.773040078105713, + "traffic_rms": 0.00018312008588175716 + }, + { + "batch_loss": 2.2993764877319336, + "innovation_rms": 6.535402715909009, + "instruction_rms": 6.53534219099546, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2749.637451171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1485233536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117374.9765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7203.404296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.210618019104004, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117257640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17153245210647583, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1652620236855e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117374.9765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7203.39697265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.210610866546631, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117257640.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.53534219099546, + "step": 261, + "teaching_rms": 6.535402715909009, + "traffic_rms": 0.0001832009786402342 + }, + { + "batch_loss": 2.322821855545044, + "innovation_rms": 5.945805835694903, + "instruction_rms": 5.945746870381875, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2743.277587890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1336710144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117363.2421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7202.68310546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.209896087646484, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117245904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24800074100494385, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8487356757977977e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117363.2421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7202.67578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2098894119262695, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117245904.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.945746870381875, + "step": 262, + "teaching_rms": 5.945805835694903, + "traffic_rms": 0.00018334431194472577 + }, + { + "batch_loss": 2.3150768280029297, + "innovation_rms": 6.245703414780859, + "instruction_rms": 6.24567305000367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2755.0927734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1203039104.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117351.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7201.96240234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.209174633026123, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117234168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21828246116638184, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.563862108218018e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117351.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7201.955078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.20916748046875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117234168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.24567305000367, + "step": 263, + "teaching_rms": 6.245703414780859, + "traffic_rms": 0.0001833844640077857 + }, + { + "batch_loss": 2.2875120639801025, + "innovation_rms": 6.590861223836605, + "instruction_rms": 6.590834473742128, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2721.67529296875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1082735232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117339.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7201.24169921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.208453178405762, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117222432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21471631526947021, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3074759155861102e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117339.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7201.234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.208446025848389, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117222432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.590834473742128, + "step": 264, + "teaching_rms": 6.590861223836605, + "traffic_rms": 0.0001835888148488851 + }, + { + "batch_loss": 2.290386915206909, + "innovation_rms": 6.237833525252429, + "instruction_rms": 6.237772677999508, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2761.047119140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 974461696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117328.0234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7200.52099609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2077317237854, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117210696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21137851476669312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.076728196698241e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117328.0234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7200.513671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.207724571228027, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117210696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.237772677999508, + "step": 265, + "teaching_rms": 6.237833525252429, + "traffic_rms": 0.00018405843812163418 + }, + { + "batch_loss": 2.321544885635376, + "innovation_rms": 6.091268040730858, + "instruction_rms": 6.091235107198598, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2773.807373046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 877015552.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117316.2890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7199.80029296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.207010746002197, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117198968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21705204248428345, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8690552678890526e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117316.2890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7199.79296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.207003116607666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117198968.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.091235107198598, + "step": 266, + "teaching_rms": 6.091268040730858, + "traffic_rms": 0.00018455262002514997 + }, + { + "batch_loss": 2.3172004222869873, + "innovation_rms": 6.345503258154754, + "instruction_rms": 6.345498499526065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2737.92333984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 789313984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117304.5546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7199.07958984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.206289291381836, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117187240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2862128019332886, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6821497411001474e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117304.5546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7199.072265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.206282138824463, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117187240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.345498499526065, + "step": 267, + "teaching_rms": 6.345503258154754, + "traffic_rms": 0.00018503892429886008 + }, + { + "batch_loss": 2.3359758853912354, + "innovation_rms": 6.131769623001766, + "instruction_rms": 6.131702812056304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2615.132080078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 710382592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117292.8203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7198.35888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.205567836761475, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117175512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5644673109054565, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5139346942305565e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117292.8203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7198.3515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.20556116104126, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117175512.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.131702812056304, + "step": 268, + "teaching_rms": 6.131769623001766, + "traffic_rms": 0.00018556591175933427 + }, + { + "batch_loss": 2.3295845985412598, + "innovation_rms": 6.631263099449841, + "instruction_rms": 6.6312473795435425, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2684.140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 639344320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117281.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7197.63818359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2048468589782715, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117163784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34454697370529175, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3625412066176068e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117281.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7197.630859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204840183258057, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117163784.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.6312473795435425, + "step": 269, + "teaching_rms": 6.631263099449841, + "traffic_rms": 0.00018586205789637293 + }, + { + "batch_loss": 2.301117181777954, + "innovation_rms": 6.227690101075762, + "instruction_rms": 6.227650446539298, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2664.101318359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 575409856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117269.359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7196.91796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204125881195068, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117152056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09984084963798523, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2262870768608991e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117269.359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7196.91064453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204118728637695, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117152056.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.227650446539298, + "step": 270, + "teaching_rms": 6.227690101075762, + "traffic_rms": 0.00018607769387552687 + }, + { + "batch_loss": 2.322810649871826, + "innovation_rms": 6.056624276410073, + "instruction_rms": 6.056589623760107, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2683.30419921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 517868864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117257.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7196.19775390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.203404903411865, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117140328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24830636382102966, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1036583600798622e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117257.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7196.1904296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.203397750854492, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117140328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.056589651144562, + "step": 271, + "teaching_rms": 6.056624276410073, + "traffic_rms": 0.00018603934526988434 + }, + { + "batch_loss": 2.298269271850586, + "innovation_rms": 5.779281463416484, + "instruction_rms": 5.779248518724946, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2651.611572265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 466081984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117245.890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7195.4775390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.202683925628662, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117128600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38974282145500183, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.9329254226177e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117245.890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7195.47021484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.202676773071289, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117128600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.779248533074251, + "step": 272, + "teaching_rms": 5.779281463416484, + "traffic_rms": 0.00018557655290730388 + }, + { + "batch_loss": 2.330063581466675, + "innovation_rms": 6.172647000330377, + "instruction_rms": 6.172606333041623, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2600.721923828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 419473792.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117234.15625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7194.75732421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.201962947845459, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117116880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25596049427986145, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.93963260750752e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117234.15625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7194.75, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.201955795288086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117116880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.172606333041623, + "step": 273, + "teaching_rms": 6.172647000330377, + "traffic_rms": 0.00018503919142657057 + }, + { + "batch_loss": 2.3256282806396484, + "innovation_rms": 6.534215474822455, + "instruction_rms": 6.53418195490701, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2618.8916015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 377526400.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117222.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7194.037109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.201242446899414, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117105160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17983432114124298, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.045669346756767e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117222.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7194.02978515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.201234817504883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117105160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.53418195490701, + "step": 274, + "teaching_rms": 6.534215474822455, + "traffic_rms": 0.0001844916916254183 + }, + { + "batch_loss": 2.3258776664733887, + "innovation_rms": 6.281332427432134, + "instruction_rms": 6.281297664956611, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2635.71044921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 339773760.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117210.6953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7193.31689453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.200521945953369, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117093440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2922641932964325, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.24110213923268e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117210.6953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7193.3095703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.20051383972168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117093440.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.281297664956611, + "step": 275, + "teaching_rms": 6.281332427432134, + "traffic_rms": 0.00018412002668095317 + }, + { + "batch_loss": 2.2988064289093018, + "innovation_rms": 6.159112411099266, + "instruction_rms": 6.159070363272192, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2647.8642578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 305796384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117198.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7192.5966796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.199801445007324, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117081720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24254827201366425, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.5169915615115315e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117198.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7192.58935546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.199792861938477, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117081720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.159070363272192, + "step": 276, + "teaching_rms": 6.159112411099266, + "traffic_rms": 0.0001838687232698029 + }, + { + "batch_loss": 2.3241872787475586, + "innovation_rms": 6.6492510911378755, + "instruction_rms": 6.649204271879292, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2665.98095703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 275216736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117187.2421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191.876953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.199080944061279, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117070000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21931470930576324, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.865292223461438e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117187.2421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191.86962890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.199072360992432, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117070000.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.649204271879292, + "step": 277, + "teaching_rms": 6.6492510911378755, + "traffic_rms": 0.0001837915131005207 + }, + { + "batch_loss": 2.3610117435455322, + "innovation_rms": 6.968398230833062, + "instruction_rms": 6.968338098633251, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2656.986083984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 247695056.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117175.515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191.1572265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.198360443115234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117058280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2171192467212677, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.278762728266884e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117175.515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191.14990234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.198351860046387, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117058280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.968338110533965, + "step": 278, + "teaching_rms": 6.968398230833062, + "traffic_rms": 0.00018383150026637677 + }, + { + "batch_loss": 2.38078236579895, + "innovation_rms": 6.260554877199166, + "instruction_rms": 6.260514680577034, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2646.7705078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 222925552.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117163.7890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7190.4375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1976399421691895, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117046560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19133231043815613, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.7508865463896655e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117163.7890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7190.43017578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.197631359100342, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117046560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.260514680577034, + "step": 279, + "teaching_rms": 6.260554877199166, + "traffic_rms": 0.00018373363462322287 + }, + { + "batch_loss": 2.331296443939209, + "innovation_rms": 6.733733564844103, + "instruction_rms": 6.733702827582524, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2627.519775390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 200632992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117152.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7189.7177734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1969194412231445, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117034848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3426187038421631, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.275797891750699e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117152.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7189.71044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.196910858154297, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117034848.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.733702827582524, + "step": 280, + "teaching_rms": 6.733733564844103, + "traffic_rms": 0.000183182475633389 + }, + { + "batch_loss": 2.3390936851501465, + "innovation_rms": 6.235762150743483, + "instruction_rms": 6.235708800722366, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2724.747314453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 180569696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117140.3359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7188.998046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1961989402771, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117023136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4386723041534424, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.848218057100894e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117140.3359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7188.99072265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.19619083404541, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117023136.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.235708800722366, + "step": 281, + "teaching_rms": 6.235762150743483, + "traffic_rms": 0.00018283499647675657 + }, + { + "batch_loss": 2.334242582321167, + "innovation_rms": 6.576739134352251, + "instruction_rms": 6.576671246191634, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2735.450439453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 162512720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117128.609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7188.2783203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.195478439331055, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117011424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26422351598739624, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.463396069491864e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117128.609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7188.27099609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.195470809936523, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 117011424.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.576671246191634, + "step": 282, + "teaching_rms": 6.576739134352251, + "traffic_rms": 0.00018249112849936898 + }, + { + "batch_loss": 2.3252670764923096, + "innovation_rms": 6.545014576345903, + "instruction_rms": 6.5449785417466915, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2655.188720703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 146261440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117116.8828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7187.55908203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.19475793838501, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116999712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21039804816246033, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.117056394330575e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117116.8828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7187.5517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.194750785827637, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116999712.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.544978529076184, + "step": 283, + "teaching_rms": 6.545014576345903, + "traffic_rms": 0.00018243360287853387 + }, + { + "batch_loss": 2.343247890472412, + "innovation_rms": 6.43744830902761, + "instruction_rms": 6.437393830680152, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2732.619140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143646656.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117105.1640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7186.83984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.194037914276123, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116988000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3506414592266083, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.80535073216015e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117105.1640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7186.83251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.19403076171875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116988000.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.437393843562416, + "step": 284, + "teaching_rms": 6.43744830902761, + "traffic_rms": 0.00018266386726521682 + }, + { + "batch_loss": 2.336399793624878, + "innovation_rms": 6.0938985089303275, + "instruction_rms": 6.093880901986062, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2625.614013671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144687552.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117093.4453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7186.12060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1933183670043945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116976288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2830529808998108, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5248154997825623e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117093.4453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7186.11328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.193310737609863, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116976288.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.093880901986062, + "step": 285, + "teaching_rms": 6.0938985089303275, + "traffic_rms": 0.00018303972446984335 + }, + { + "batch_loss": 2.3309030532836914, + "innovation_rms": 6.177533150096755, + "instruction_rms": 6.177501431438714, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2563.67919921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145593872.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117081.734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7185.4013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.192598342895508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116964576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3046990633010864, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.272333858854836e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117081.734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7185.39404296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.192590713500977, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116964576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.177501431438714, + "step": 286, + "teaching_rms": 6.177533150096755, + "traffic_rms": 0.00018350454425025275 + }, + { + "batch_loss": 2.3260931968688965, + "innovation_rms": 6.407042451430201, + "instruction_rms": 6.406996207036885, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2615.864013671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 146281008.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117070.015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7184.68212890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.191878318786621, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116952872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34952273964881897, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0451004729693523e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117070.015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7184.6748046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.19187068939209, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116952872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.406996207036885, + "step": 287, + "teaching_rms": 6.407042451430201, + "traffic_rms": 0.00018404243179179596 + }, + { + "batch_loss": 2.2892401218414307, + "innovation_rms": 6.731932072226192, + "instruction_rms": 6.731875332122377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2646.594482421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145846864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117058.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7183.962890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.191158294677734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116941168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17946496605873108, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8405903574603144e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117058.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7183.95556640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.191150665283203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116941168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.731875344441114, + "step": 288, + "teaching_rms": 6.731932072226192, + "traffic_rms": 0.0001845587532713901 + }, + { + "batch_loss": 2.3030292987823486, + "innovation_rms": 6.545621740452894, + "instruction_rms": 6.545558407119962, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2689.61962890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145454448.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117046.578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7183.24365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.190438270568848, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116929464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16820679605007172, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6565312535021803e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117046.578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7183.236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.190431118011475, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116929464.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.545558419789348, + "step": 289, + "teaching_rms": 6.545621740452894, + "traffic_rms": 0.00018505212662449548 + }, + { + "batch_loss": 2.294813632965088, + "innovation_rms": 6.1487046442292845, + "instruction_rms": 6.148656835071432, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2753.789306640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145016272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117034.859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7182.52490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.189718246459961, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116917760.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11494273692369461, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.490878048571176e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117034.859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7182.517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.189711570739746, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116917760.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.148656835071432, + "step": 290, + "teaching_rms": 6.1487046442292845, + "traffic_rms": 0.00018533335883640656 + }, + { + "batch_loss": 2.3114736080169678, + "innovation_rms": 6.69798369697785, + "instruction_rms": 6.697963428159754, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2788.370849609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144255200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117023.140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7181.80615234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.188998699188232, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116906056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.150255486369133, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3417901527645881e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117023.140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7181.798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.188992023468018, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116906056.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.697963428159754, + "step": 291, + "teaching_rms": 6.69798369697785, + "traffic_rms": 0.00018520690489361345 + }, + { + "batch_loss": 2.323643445968628, + "innovation_rms": 6.478099201072765, + "instruction_rms": 6.478094669801708, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2770.1181640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144464640.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117011.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7181.08740234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.188279151916504, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116894352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2482614368200302, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.207611148856813e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117011.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7181.080078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.188272476196289, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116894352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.478094669801708, + "step": 292, + "teaching_rms": 6.478099201072765, + "traffic_rms": 0.0001851100868447983 + }, + { + "batch_loss": 2.3325355052948, + "innovation_rms": 6.4906590186707955, + "instruction_rms": 6.4906785927554544, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2782.699462890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144174416.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116999.7109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7180.36865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.187559604644775, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116882656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4092676043510437, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0868500339711318e-06, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116999.7109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7180.361328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.187553405761719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116882656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.490678605531961, + "step": 293, + "teaching_rms": 6.4906590186707955, + "traffic_rms": 0.0001849476909850132 + }, + { + "batch_loss": 2.3259315490722656, + "innovation_rms": 6.896209899405662, + "instruction_rms": 6.896180701315592, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2830.95947265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142720608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116988.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7179.64990234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.186840057373047, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116870960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2768603563308716, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.78164962361916e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116988.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7179.642578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.18683385848999, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116870960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.896180701315592, + "step": 294, + "teaching_rms": 6.896209899405662, + "traffic_rms": 0.00018498342015245466 + }, + { + "batch_loss": 2.3205902576446533, + "innovation_rms": 6.283666568635043, + "instruction_rms": 6.283618347293503, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2838.888427734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143827040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116976.2890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7178.93115234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.186120986938477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116859264.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12692809104919434, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.803484661257244e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116976.2890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7178.923828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.186114311218262, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116859264.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.283618347293503, + "step": 295, + "teaching_rms": 6.283666568635043, + "traffic_rms": 0.0001850675982575969 + }, + { + "batch_loss": 2.3109331130981445, + "innovation_rms": 6.361617124296339, + "instruction_rms": 6.3615916514837, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2878.19677734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143472544.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116964.578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7178.21240234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.185401916503906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116847568.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24292975664138794, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.923135854071006e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116964.578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7178.205078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.185394763946533, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116847568.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.361591664519462, + "step": 296, + "teaching_rms": 6.361617124296339, + "traffic_rms": 0.0001849637058276959 + }, + { + "batch_loss": 2.3413121700286865, + "innovation_rms": 5.988715203027085, + "instruction_rms": 5.988639237374529, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2851.441162109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143829424.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116952.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7177.494140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.184682846069336, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116835872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5254322290420532, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.130822154977068e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116952.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7177.48681640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.184675693511963, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116835872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.988639237374529, + "step": 297, + "teaching_rms": 5.988715203027085, + "traffic_rms": 0.00018484894558179753 + }, + { + "batch_loss": 2.299436569213867, + "innovation_rms": 6.381788201402707, + "instruction_rms": 6.381759353769093, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2814.921142578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145237616.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116941.171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7176.77587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.183963775634766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116824176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5286310911178589, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.417739655262267e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116941.171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7176.7685546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.183956623077393, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116824176.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.381759353769093, + "step": 298, + "teaching_rms": 6.381788201402707, + "traffic_rms": 0.00018447449704220086 + }, + { + "batch_loss": 2.346522808074951, + "innovation_rms": 6.740169649841865, + "instruction_rms": 6.740144096112867, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2792.87451171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145148752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116929.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7176.0576171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.183244705200195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116812480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24813365936279297, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.775965519205783e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116929.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7176.05029296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.183237552642822, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116812480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.740144108416492, + "step": 299, + "teaching_rms": 6.740169649841865, + "traffic_rms": 0.00018436187208056522 + }, + { + "batch_loss": 2.348090887069702, + "innovation_rms": 6.11466636311055, + "instruction_rms": 6.114681351407388, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2750.519775390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144521520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116917.765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7175.33935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.182525634765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116800792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29667702317237854, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.19836873991153e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116917.765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7175.33203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.18251895904541, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116800792.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.114681351407388, + "step": 300, + "teaching_rms": 6.11466636311055, + "traffic_rms": 0.00018434817120636447 + }, + { + "batch_loss": 2.327596664428711, + "innovation_rms": 6.366644247859935, + "instruction_rms": 6.366600965876852, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2697.64501953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144357360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116906.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7174.62109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.181807041168213, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116789104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37959352135658264, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.67853169539012e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116906.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7174.61376953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.18179988861084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116789104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.366600965876852, + "step": 301, + "teaching_rms": 6.366644247859935, + "traffic_rms": 0.00018419846101159796 + }, + { + "batch_loss": 2.3213629722595215, + "innovation_rms": 6.465648772117037, + "instruction_rms": 6.4655954178323105, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2605.37646484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144606752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116894.359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7173.90283203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.181088447570801, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116777416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6371521353721619, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.210678525851108e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116894.359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7173.8955078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1810808181762695, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116777416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.4655954178323105, + "step": 302, + "teaching_rms": 6.465648772117037, + "traffic_rms": 0.00018403753024628504 + }, + { + "batch_loss": 2.337709665298462, + "innovation_rms": 6.448871983431422, + "instruction_rms": 6.448792089968623, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2586.149658203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145233696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116882.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7173.1845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1803693771362305, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116765728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6531166434288025, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7896106164225785e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116882.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7173.17724609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.180362224578857, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116765728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.448792089968623, + "step": 303, + "teaching_rms": 6.448871983431422, + "traffic_rms": 0.00018386149837554047 + }, + { + "batch_loss": 2.3156936168670654, + "innovation_rms": 6.584645014559821, + "instruction_rms": 6.5845969169234815, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2587.82666015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144894832.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116870.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7172.466796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.179650783538818, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116754040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3146752119064331, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4106494695151923e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116870.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7172.45947265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.179643630981445, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116754040.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.584596929517753, + "step": 304, + "teaching_rms": 6.584645014559821, + "traffic_rms": 0.00018368120076853845 + }, + { + "batch_loss": 2.310894250869751, + "innovation_rms": 6.344575615656818, + "instruction_rms": 6.344503440179632, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2575.609130859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144597984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116859.2734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7171.7490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.178932189941406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116742352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30503374338150024, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0695844088768354e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116859.2734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7171.74169921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.178925037384033, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116742352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.344503440179632, + "step": 305, + "teaching_rms": 6.344575615656818, + "traffic_rms": 0.00018374029288733588 + }, + { + "batch_loss": 2.3276708126068115, + "innovation_rms": 6.15570506715524, + "instruction_rms": 6.15565058779062, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2593.840576171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144401632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116847.578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7171.03125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.178213596343994, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116730664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3796680271625519, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.762625967989152e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116847.578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7171.02392578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.178206443786621, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116730664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.15565058779062, + "step": 306, + "teaching_rms": 6.15570506715524, + "traffic_rms": 0.0001838344108956026 + }, + { + "batch_loss": 2.327651262283325, + "innovation_rms": 6.759254123176603, + "instruction_rms": 6.759213734275597, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2594.09130859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145045168.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116835.890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7170.3134765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.177495002746582, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116718984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5988670587539673, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.486363257503399e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116835.890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7170.30615234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.177488327026367, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116718984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.759213734275597, + "step": 307, + "teaching_rms": 6.759254123176603, + "traffic_rms": 0.00018392080131831326 + }, + { + "batch_loss": 2.3277695178985596, + "innovation_rms": 6.80147884292368, + "instruction_rms": 6.801440449833168, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2563.060546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 146928672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116824.1953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7169.595703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.176776885986328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116707304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.540228009223938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2377268749096402e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116824.1953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7169.58837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.176769733428955, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116707304.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.801440449833168, + "step": 308, + "teaching_rms": 6.80147884292368, + "traffic_rms": 0.0001841879616454497 + }, + { + "batch_loss": 2.3205454349517822, + "innovation_rms": 6.332342850944266, + "instruction_rms": 6.33229422841313, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2584.722412109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 147177344.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116812.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7168.8779296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.176058292388916, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116695624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2735595703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0139540879426931e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116812.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7168.87060546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.176051616668701, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116695624.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.33229422841313, + "step": 309, + "teaching_rms": 6.332342850944266, + "traffic_rms": 0.00018435293043170422 + }, + { + "batch_loss": 2.33591890335083, + "innovation_rms": 6.307061193430264, + "instruction_rms": 6.307035753210692, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2666.6572265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 146462976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116800.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7168.16015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.175340175628662, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116683944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4854196310043335, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.812558565461586e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116800.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7168.15283203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.175333499908447, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116683944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.307035753210692, + "step": 310, + "teaching_rms": 6.307061193430264, + "traffic_rms": 0.0001842310823801205 + }, + { + "batch_loss": 2.3357276916503906, + "innovation_rms": 6.066072702640255, + "instruction_rms": 6.0660237917106565, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2658.528076171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 146894720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116789.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7167.44287109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.174622058868408, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116672264.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5296362042427063, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6313026662828634e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116789.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7167.435546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.174615383148193, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116672264.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.0660237917106565, + "step": 311, + "teaching_rms": 6.066072702640255, + "traffic_rms": 0.00018406650789938448 + }, + { + "batch_loss": 2.3259575366973877, + "innovation_rms": 6.301446072885848, + "instruction_rms": 6.301389998877578, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2690.56396484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 146878848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116777.4375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7166.7255859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.173903942108154, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116660584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34781694412231445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4681724280762865e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116777.4375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7166.71826171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1738972663879395, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116660584.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.301389998877578, + "step": 312, + "teaching_rms": 6.301446072885848, + "traffic_rms": 0.00018381602543139378 + }, + { + "batch_loss": 2.344186544418335, + "innovation_rms": 6.813966994186366, + "instruction_rms": 6.813930663699016, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2741.58349609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 145433104.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116765.7421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7166.00830078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1731858253479, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116648904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26324132084846497, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3213551142143842e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116765.7421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7166.0009765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1731791496276855, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116648904.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.813930663699016, + "step": 313, + "teaching_rms": 6.813966994186366, + "traffic_rms": 0.00018367509876928507 + }, + { + "batch_loss": 2.318521022796631, + "innovation_rms": 5.904570979765386, + "instruction_rms": 5.904497878792977, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2792.561767578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 144167088.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116754.0546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7165.291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.172468185424805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116637232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42070573568344116, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1892196027929458e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116754.0546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7165.28369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.172461032867432, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116637232.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.904497878792977, + "step": 314, + "teaching_rms": 5.904570979765386, + "traffic_rms": 0.00018382627042878716 + }, + { + "batch_loss": 2.325622320175171, + "innovation_rms": 6.104456355435098, + "instruction_rms": 6.104446913986636, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2747.30517578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143979808.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116742.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7164.57373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.171750068664551, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116625560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3886701762676239, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0702976283027965e-07, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116742.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7164.56640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.171743392944336, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116625560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.10444692757152, + "step": 315, + "teaching_rms": 6.104456355435098, + "traffic_rms": 0.00018378450705584294 + }, + { + "batch_loss": 2.3041412830352783, + "innovation_rms": 6.355309626503075, + "instruction_rms": 6.355277266227254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2820.94775390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143613648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116730.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7163.8564453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.171032428741455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116613888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41272062063217163, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.632678654725169e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116730.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7163.84912109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.171025276184082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116613888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.355277279275969, + "step": 316, + "teaching_rms": 6.355309626503075, + "traffic_rms": 0.00018396048497002206 + }, + { + "batch_loss": 2.325312852859497, + "innovation_rms": 6.545011991807075, + "instruction_rms": 6.544996875457787, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2833.962158203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143479344.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116719.0078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7163.1396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.170314788818359, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116602216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2658461928367615, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.669410789252652e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116719.0078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7163.13232421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.170307636260986, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116602216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.544996875457787, + "step": 317, + "teaching_rms": 6.545011991807075, + "traffic_rms": 0.0001841759788899151 + }, + { + "batch_loss": 2.303226947784424, + "innovation_rms": 6.479358639604041, + "instruction_rms": 6.479330866215064, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2819.48583984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142491056.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116707.3203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7162.4228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.169597148895264, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116590544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2650202512741089, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.80246978138166e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116707.3203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7162.41552734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.169589996337891, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116590544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.479330879013948, + "step": 318, + "teaching_rms": 6.479358639604041, + "traffic_rms": 0.00018458602901787335 + }, + { + "batch_loss": 2.3050684928894043, + "innovation_rms": 6.475160440121357, + "instruction_rms": 6.475133738372171, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2901.937744140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141781136.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116695.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7161.7060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.168879508972168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116578872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3020574748516083, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.022222803243494e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116695.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7161.69873046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.168872356414795, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116578872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.47513376398653, + "step": 319, + "teaching_rms": 6.475160440121357, + "traffic_rms": 0.00018499609206042457 + }, + { + "batch_loss": 2.298576831817627, + "innovation_rms": 6.678851216126064, + "instruction_rms": 6.6788278107763075, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2904.374755859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141300752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116683.953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7160.9892578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1681623458862305, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116567200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34833449125289917, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.320000522919145e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116683.953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7160.98193359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.168154716491699, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116567200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.678827823192887, + "step": 320, + "teaching_rms": 6.678851216126064, + "traffic_rms": 0.00018528513923406 + }, + { + "batch_loss": 2.350660800933838, + "innovation_rms": 7.066900199936269, + "instruction_rms": 7.066853519076067, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2947.692626953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140994800.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116672.2734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7160.2724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.167444705963135, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116555536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3002622127532959, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.68800047062723e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116672.2734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7160.26513671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.167437553405762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116555536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7.066853530810879, + "step": 321, + "teaching_rms": 7.066900199936269, + "traffic_rms": 0.00018554835719289186 + }, + { + "batch_loss": 2.3076624870300293, + "innovation_rms": 6.4692145621922945, + "instruction_rms": 6.469177939423824, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2933.1748046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141080720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116660.6015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7159.5556640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.166727542877197, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116543872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31256192922592163, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.1192003525102336e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116660.6015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7159.54833984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.166719913482666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116543872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.469177939423824, + "step": 322, + "teaching_rms": 6.4692145621922945, + "traffic_rms": 0.0001856964733500433 + }, + { + "batch_loss": 2.339111089706421, + "innovation_rms": 6.913233960284256, + "instruction_rms": 6.913166293449238, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2968.48486328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141001936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116648.921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7158.8388671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.16601037979126, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116532208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19155248999595642, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.607280246204937e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116648.921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7158.83154296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1660027503967285, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116532208.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.913166293449238, + "step": 323, + "teaching_rms": 6.913233960284256, + "traffic_rms": 0.0001856071677759874 + }, + { + "batch_loss": 2.3245480060577393, + "innovation_rms": 6.35560073220243, + "instruction_rms": 6.355521233130861, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3006.855224609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141421568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116637.2421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7158.12255859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.165292739868164, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116520544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36372435092926025, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.14655225711158e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116637.2421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7158.115234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.165285587310791, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116520544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.355521246179075, + "step": 324, + "teaching_rms": 6.35560073220243, + "traffic_rms": 0.0001851857530002812 + }, + { + "batch_loss": 2.287607192993164, + "innovation_rms": 6.564856332573171, + "instruction_rms": 6.5648314731548325, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2951.76806640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140424128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116625.5703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7157.40625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.164575576782227, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116508880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2889968454837799, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7318969248190115e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116625.5703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7157.39892578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.164567947387695, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116508880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.5648314731548325, + "step": 325, + "teaching_rms": 6.564856332573171, + "traffic_rms": 0.00018435865474561696 + }, + { + "batch_loss": 2.3452048301696777, + "innovation_rms": 5.773760253950098, + "instruction_rms": 5.773735913378582, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2934.713623046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140069456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116613.8984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7156.68994140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.163858413696289, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116497216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21735242009162903, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.358707090228563e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116613.8984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7156.6826171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.163850784301758, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116497216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.773735913378582, + "step": 326, + "teaching_rms": 5.773760253950098, + "traffic_rms": 0.0001837211043264552 + }, + { + "batch_loss": 2.3138463497161865, + "innovation_rms": 6.611616518350881, + "instruction_rms": 6.611531128124851, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2969.645263671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 139245168.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116602.2265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7155.9736328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.16314172744751, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116485552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34114205837249756, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.022836381205707e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116602.2265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7155.96630859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.16313362121582, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116485552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.611531128124851, + "step": 327, + "teaching_rms": 6.611616518350881, + "traffic_rms": 0.00018290788802044955 + }, + { + "batch_loss": 2.324497938156128, + "innovation_rms": 6.479210374004736, + "instruction_rms": 6.479192597889245, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2881.775634765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 139984768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116590.5546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7155.25732421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.162424564361572, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116473896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34363433718681335, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7205526720308626e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116590.5546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7155.25, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.162416934967041, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116473896.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.479192597889245, + "step": 328, + "teaching_rms": 6.479210374004736, + "traffic_rms": 0.0001821935522459808 + }, + { + "batch_loss": 2.3420169353485107, + "innovation_rms": 6.356045682669413, + "instruction_rms": 6.356043909470828, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2820.898193359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140687376.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116578.890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7154.541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.161707401275635, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116462240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42481690645217896, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4484974048277763e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116578.890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7154.53369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.161700248718262, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116462240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.356043909470828, + "step": 329, + "teaching_rms": 6.356045682669413, + "traffic_rms": 0.00018192870007981723 + }, + { + "batch_loss": 2.300502061843872, + "innovation_rms": 6.558503788813574, + "instruction_rms": 6.5584482420372945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2790.840576171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141233904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116567.2265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7153.82470703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1609907150268555, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116450584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26789748668670654, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2036475755271567e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116567.2265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7153.8173828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.160983085632324, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116450584.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.5584482420372945, + "step": 330, + "teaching_rms": 6.558503788813574, + "traffic_rms": 0.00018233955473301346 + }, + { + "batch_loss": 2.3220653533935547, + "innovation_rms": 6.604433969730207, + "instruction_rms": 6.604379613266801, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2848.843017578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140549472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116555.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7153.10888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.160274028778076, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116438928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22979636490345, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9832826936294623e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116555.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7153.1015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.160266399383545, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116438928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.604379625823348, + "step": 331, + "teaching_rms": 6.604433969730207, + "traffic_rms": 0.0001831618388324071 + }, + { + "batch_loss": 2.3120439052581787, + "innovation_rms": 6.088726755817639, + "instruction_rms": 6.0886654550706165, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2876.693603515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140614432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116543.890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7152.39306640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.159557342529297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116427272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2349226027727127, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7849544065029477e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116543.890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7152.3857421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.159549713134766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116427272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.0886654550706165, + "step": 332, + "teaching_rms": 6.088726755817639, + "traffic_rms": 0.00018411420769524867 + }, + { + "batch_loss": 2.32232928276062, + "innovation_rms": 6.152706045646764, + "instruction_rms": 6.152674345494996, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2949.6015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 139740832.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116532.2265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7151.67724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.158841133117676, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116415616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29092228412628174, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.606458965852653e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116532.2265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7151.669921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.158833026885986, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116415616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.152674358973396, + "step": 333, + "teaching_rms": 6.152706045646764, + "traffic_rms": 0.00018478275162488043 + }, + { + "batch_loss": 2.323047637939453, + "innovation_rms": 6.479721628842034, + "instruction_rms": 6.479691835516988, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3000.06201171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 139874944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116520.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7150.96142578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1581244468688965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116403960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20708055794239044, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4458130692673876e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116520.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7150.9541015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.158116817474365, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116403960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.479691835516988, + "step": 334, + "teaching_rms": 6.479721628842034, + "traffic_rms": 0.00018526875865194288 + }, + { + "batch_loss": 2.3226511478424072, + "innovation_rms": 6.285045207581321, + "instruction_rms": 6.285033095815178, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3026.640380859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140359296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116508.8984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7150.24560546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.157407760620117, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116392312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19446714222431183, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3012317445770805e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116508.8984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7150.23828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.157400608062744, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116392312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.285033095815178, + "step": 335, + "teaching_rms": 6.285045207581321, + "traffic_rms": 0.00018560060977438757 + }, + { + "batch_loss": 2.298741579055786, + "innovation_rms": 5.743622772566537, + "instruction_rms": 5.743583329278729, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2981.072998046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141105984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116497.234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7149.52978515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.156691074371338, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116380664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22454802691936493, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1711085434740198e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116497.234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7149.5224609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.156684398651123, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116380664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.743583343717137, + "step": 336, + "teaching_rms": 5.743622772566537, + "traffic_rms": 0.0001857184986048867 + }, + { + "batch_loss": 2.3137261867523193, + "innovation_rms": 6.343408724354852, + "instruction_rms": 6.3433961480240395, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2967.216064453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 140432784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116485.578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7148.81396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.155974864959717, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116369016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22917652130126953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0539976891266178e-08, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116485.578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7148.806640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.155967712402344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116369016.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.3433961480240395, + "step": 337, + "teaching_rms": 6.343408724354852, + "traffic_rms": 0.00018553586953098136 + }, + { + "batch_loss": 2.2894296646118164, + "innovation_rms": 5.936102060565386, + "instruction_rms": 5.936020922567239, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2836.674560546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141518896.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116473.9140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7148.0986328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.155258655548096, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116357368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14356592297554016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.485979113321719e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116473.9140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7148.09130859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1552510261535645, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116357368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.936020922567239, + "step": 338, + "teaching_rms": 5.936102060565386, + "traffic_rms": 0.00018532520419676498 + }, + { + "batch_loss": 2.323439121246338, + "innovation_rms": 6.588244718726057, + "instruction_rms": 6.588213476266131, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2801.874755859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142369840.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116462.2578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7147.38330078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.154542446136475, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116345720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20378348231315613, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.537380580264653e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116462.2578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7147.3759765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.154534816741943, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116345720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.588213476266131, + "step": 339, + "teaching_rms": 6.588244718726057, + "traffic_rms": 0.0001847881817592601 + }, + { + "batch_loss": 2.318883180618286, + "innovation_rms": 6.287715926368904, + "instruction_rms": 6.287673141137641, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2801.969970703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142508960.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116450.6015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7146.66796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1538262367248535, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116334072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28921782970428467, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.68364216696682e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116450.6015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7146.66064453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.153818607330322, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116334072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.287673141137641, + "step": 340, + "teaching_rms": 6.287715926368904, + "traffic_rms": 0.00018432387713337902 + }, + { + "batch_loss": 2.307809829711914, + "innovation_rms": 6.74076766634721, + "instruction_rms": 6.740739109583374, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2776.56591796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142563072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116438.9453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7145.95263671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.153110504150391, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116322432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3145782947540283, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.915277683816612e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116438.9453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7145.9453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.153102397918701, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116322432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.740739097280835, + "step": 341, + "teaching_rms": 6.74076766634721, + "traffic_rms": 0.00018395899371284095 + }, + { + "batch_loss": 2.314749002456665, + "innovation_rms": 6.486690780641876, + "instruction_rms": 6.48667240867232, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2729.28662109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143414384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116427.2890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7145.2373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.152394771575928, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116310792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35417720675468445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.223749959843872e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116427.2890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7145.22998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.15238618850708, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116310792.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.48667240867232, + "step": 342, + "teaching_rms": 6.486690780641876, + "traffic_rms": 0.00018417770992965163 + }, + { + "batch_loss": 2.3250651359558105, + "innovation_rms": 6.358091888104337, + "instruction_rms": 6.358069090306716, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2805.83740234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142752896.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116415.6328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7144.52197265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.151679039001465, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116299152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2345060110092163, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.6013749194505635e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116415.6328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7144.5146484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.151670455932617, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116299152.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.358069090306716, + "step": 343, + "teaching_rms": 6.358091888104337, + "traffic_rms": 0.00018455758154249828 + }, + { + "batch_loss": 2.314680814743042, + "innovation_rms": 6.543731792883309, + "instruction_rms": 6.543686122140339, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2881.500732421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142269072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116403.984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7143.806640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.150963306427002, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116287512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2810690701007843, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.041237205460902e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116403.984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7143.79931640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.150954723358154, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116287512.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.543686122140339, + "step": 344, + "teaching_rms": 6.543731792883309, + "traffic_rms": 0.00018484695342285234 + }, + { + "batch_loss": 2.3138134479522705, + "innovation_rms": 6.26735784601638, + "instruction_rms": 6.267305331729926, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2846.5517578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142115488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116392.3359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7143.091796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.150247573852539, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116275872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3674173951148987, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.537113351688049e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116392.3359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7143.08447265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.150238990783691, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116275872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.267305331729926, + "step": 345, + "teaching_rms": 6.26735784601638, + "traffic_rms": 0.0001847824573765879 + }, + { + "batch_loss": 2.2761194705963135, + "innovation_rms": 6.860400877303943, + "instruction_rms": 6.860353866735314, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2808.802734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142215104.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116380.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7142.376953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.149531841278076, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116264232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3311331570148468, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.083402060928165e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116380.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7142.36962890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1495232582092285, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116264232.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.860353866735314, + "step": 346, + "teaching_rms": 6.860400877303943, + "traffic_rms": 0.00018449964198749075 + }, + { + "batch_loss": 2.3424220085144043, + "innovation_rms": 6.364358942199689, + "instruction_rms": 6.364282627927493, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2879.73876953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142010640.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116369.0390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7141.662109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.148816108703613, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116252592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3355037271976471, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6750618104264277e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116369.0390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7141.65478515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.148807525634766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116252592.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.364282627927493, + "step": 347, + "teaching_rms": 6.364358942199689, + "traffic_rms": 0.00018436253564891404 + }, + { + "batch_loss": 2.3175699710845947, + "innovation_rms": 6.849707228627267, + "instruction_rms": 6.849647880420544, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2861.4619140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143031168.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116357.390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7140.947265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.14810037612915, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116240960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4275456368923187, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3075555627704034e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116357.390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7140.93994140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.148092269897461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116240960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.849647880420544, + "step": 348, + "teaching_rms": 6.849707228627267, + "traffic_rms": 0.00018383896160229518 + }, + { + "batch_loss": 2.2963449954986572, + "innovation_rms": 6.328822897850293, + "instruction_rms": 6.3287735203287205, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2850.9892578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142596176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116345.7421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7140.232421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1473846435546875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116229328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3679439425468445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.976799917675521e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116345.7421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7140.22509765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.147377014160156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116229328.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.3287735203287205, + "step": 349, + "teaching_rms": 6.328822897850293, + "traffic_rms": 0.00018356787367988456 + }, + { + "batch_loss": 2.318504571914673, + "innovation_rms": 6.247094727554016, + "instruction_rms": 6.247075640835653, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2807.304443359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143035200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116334.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7139.517578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.146668910980225, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116217696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12766197323799133, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6791198148856665e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116334.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7139.51025390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.146661758422852, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116217696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.247075640835653, + "step": 350, + "teaching_rms": 6.247094727554016, + "traffic_rms": 0.00018333007667527357 + }, + { + "batch_loss": 2.323054075241089, + "innovation_rms": 5.961981093453792, + "instruction_rms": 5.961940384054717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2763.046630859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142878528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116322.453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7138.80322265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.14595365524292, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116206064.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.169703871011734, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4112076779658764e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116322.453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7138.7958984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.145946502685547, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116206064.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.961940384054717, + "step": 351, + "teaching_rms": 5.961981093453792, + "traffic_rms": 0.0001831780300463299 + }, + { + "batch_loss": 2.327579975128174, + "innovation_rms": 6.235399441812911, + "instruction_rms": 6.235367485130701, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2841.102783203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 143663712.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116310.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7138.0888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.145238876342773, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116194432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 494176.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40935218334198, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 649938.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.170086776942526e-09, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116310.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7138.08154296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.145231246948242, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116194432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11661697179079056, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14589270949363708, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.235367485130701, + "step": 352, + "teaching_rms": 6.235399441812911, + "traffic_rms": 0.00018287902183983132 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} -- cgit v1.2.3