diff options
Diffstat (limited to 'results/kp_innovation_stability_grid/margin_0p001.json')
| -rw-r--r-- | results/kp_innovation_stability_grid/margin_0p001.json | 33199 |
1 files changed, 33199 insertions, 0 deletions
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 +} |