diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 16:54:12 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 16:54:12 -0500 |
| commit | 7954dc472683aad665d05b00ab3d9f32902e140c (patch) | |
| tree | fb4dfe88b0985c6a48f365f78cecf69beafb4bd1 /results/kp_innovation_stability/closed_form_frozen_innovation.json | |
| parent | 0ae690e7b3144da2c5b3384cd7169e86529fd26f (diff) | |
diagnostic: record pre-grid stability candidates
Diffstat (limited to 'results/kp_innovation_stability/closed_form_frozen_innovation.json')
| -rw-r--r-- | results/kp_innovation_stability/closed_form_frozen_innovation.json | 34611 |
1 files changed, 34611 insertions, 0 deletions
diff --git a/results/kp_innovation_stability/closed_form_frozen_innovation.json b/results/kp_innovation_stability/closed_form_frozen_innovation.json new file mode 100644 index 0000000..d170d1b --- /dev/null +++ b/results/kp_innovation_stability/closed_form_frozen_innovation.json @@ -0,0 +1,34611 @@ +{ + "first_any_nonfinite_groups": { + "bn_running_var": [ + 4 + ] + }, + "first_any_nonfinite_step": 189, + "first_nonfinite_by_group": { + "bn_running_var": { + "indices": [ + 4 + ], + "step": 189 + } + }, + "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": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 1.863524978773512e-08, + "mse": 1.4794731586767108e-18, + "observations": 64, + "residual_traffic_rms_ratio": 7.214966198849327e-08 + }, + "first_mse": 1.4794731586767108e-18, + "last_mse": 1.4794731586767108e-18, + "post_warmup_traffic_residual_rms_ratio": 7.214966198849327e-08, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_mixed_traffic_nonfinite_diagnosis_v1", + "provenance": { + "git_commit": "c8dd2e591c9835b8618fc34bc1d043634ce345e1", + "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" + }, + "test_evaluations": 0, + "traffic_calibration": { + "instruction_rms": [ + 0.008471463806927204, + 0.006617916747927666, + 0.003988116513937712, + 0.0043865954503417015, + 0.002713605063036084, + 0.003158062230795622, + 0.002004207344725728, + 0.003757704282179475, + 0.0027085731271654367, + 0.0030200635083019733, + 0.002056022873148322, + 0.002389939734712243, + 0.0016842411132529378, + 0.0034010226372629404, + 0.002549725119024515, + 0.0030542907770723104, + 0.0020918569061905146, + 0.002605808200314641, + 0.0018916188273578882 + ], + "realized_traffic_instruction_rms_ratio": [ + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 3.999999523162842, + 4.0, + 4.000000476837158, + 3.999999761581421, + 4.0, + 4.0, + 3.999999761581421, + 4.0, + 4.000000476837158, + 4.000000476837158, + 4.0, + 4.0, + 4.0 + ], + "target_ratio": 4.0, + "traffic_gain": [ + 0.0443243570625782, + 0.035817068070173264, + 0.013778294436633587, + 0.022762222215533257, + 0.007214079145342112, + 0.017218582332134247, + 0.004616298712790012, + 0.02042512409389019, + 0.007327491417527199, + 0.01611751690506935, + 0.004703814629465342, + 0.01256294921040535, + 0.003324519144371152, + 0.01805141195654869, + 0.006063441745936871, + 0.016296451911330223, + 0.0044285994954407215, + 0.013874167576432228, + 0.0036007820162922144 + ], + "unscaled_traffic_rms": [ + 0.7644973993301392, + 0.7390796542167664, + 1.157796859741211, + 0.7708553671836853, + 1.504616141319275, + 0.7336404919624329, + 1.736635684967041, + 0.7358984351158142, + 1.478581428527832, + 0.7495108246803284, + 1.7483876943588257, + 0.7609485983848572, + 2.0264477729797363, + 0.7536302804946899, + 1.6820316314697266, + 0.7496824264526367, + 1.8894071578979492, + 0.7512690424919128, + 2.10134220123291 + ] + }, + "trajectory": [ + { + "batch_loss": 2.9134020805358887, + "innovation_rms": 0.004215560388414443, + "instruction_rms": 0.004215397885545187, + "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": 0.053474150598049164, + "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": 0.00811514537781477, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08115145564079285, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0037773847579956, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03945024311542511, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05994537100195885, + "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": 1.1170727014541626, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017318261937564094, + "step": 1, + "teaching_rms": 0.004215560388414443, + "traffic_rms": 0.016800931977973527 + }, + { + "batch_loss": 2.406020402908325, + "innovation_rms": 0.004092005873095015, + "instruction_rms": 0.004091849582943296, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7660105228424072, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.962428331375122, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.055049605667591095, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4140651524066925, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3981689214706421, + "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": 0.014180497266352177, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.062249816954135895, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.008810043334961, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05032625049352646, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08590418100357056, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3788767158985138, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39816346764564514, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1178290843963623, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017319218796219168, + "step": 2, + "teaching_rms": 0.004092005873095015, + "traffic_rms": 0.016830574970372936 + }, + { + "batch_loss": 2.6357274055480957, + "innovation_rms": 0.0042798609914381095, + "instruction_rms": 0.00427967987849893, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0908468961715698, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8036653995513916, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06017614156007767, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4310041069984436, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3125629425048828, + "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": 0.017004597932100296, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05380966514348984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0125212669372559, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058248236775398254, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12273018062114716, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3962387144565582, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3126381039619446, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1186535358428955, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017367473853653646, + "step": 3, + "teaching_rms": 0.0042798609914381095, + "traffic_rms": 0.016834298035533213 + }, + { + "batch_loss": 2.8387794494628906, + "innovation_rms": 0.004146550494775722, + "instruction_rms": 0.004146363287045007, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3724877834320068, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.761442184448242, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07205741852521896, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44220468401908875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23472772538661957, + "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": 0.01920175924897194, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07674350589513779, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0174511671066284, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08731257915496826, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11141785234212875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39774709939956665, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23471370339393616, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1190940141677856, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017323035576264804, + "step": 4, + "teaching_rms": 0.004146550494775722, + "traffic_rms": 0.016820016436575137 + }, + { + "batch_loss": 2.7710607051849365, + "innovation_rms": 0.004113651093429867, + "instruction_rms": 0.004113474630717812, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.627219796180725, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.652846813201904, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07735639810562134, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4483832120895386, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4076094627380371, + "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": 0.027102595195174217, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08502216637134552, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.023786187171936, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09597021341323853, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12033584713935852, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3851354718208313, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4075927734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1193541288375854, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017357861116620073, + "step": 5, + "teaching_rms": 0.004113651093429867, + "traffic_rms": 0.016867746899459607 + }, + { + "batch_loss": 2.694866180419922, + "innovation_rms": 0.004048320988413004, + "instruction_rms": 0.004047927237824672, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85043203830719, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.51856803894043, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08964185416698456, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45894479751586914, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4642164707183838, + "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": 0.035817164927721024, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08808646351099014, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0332908630371094, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09611919522285461, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11808763444423676, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39148005843162537, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.464184045791626, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1196045875549316, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017348484973496427, + "step": 6, + "teaching_rms": 0.004048320988413004, + "traffic_rms": 0.016871919583135386 + }, + { + "batch_loss": 2.5539674758911133, + "innovation_rms": 0.004105014702470099, + "instruction_rms": 0.0041044846961079805, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1285974979400635, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.206201553344727, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1302187591791153, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4740625321865082, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6080023646354675, + "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": 0.04346618801355362, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09067319333553314, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.044256567955017, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10965670645236969, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13024958968162537, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3989025950431824, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6079418063163757, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1193941831588745, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017379491530606167, + "step": 7, + "teaching_rms": 0.004105014702470099, + "traffic_rms": 0.016882168804390293 + }, + { + "batch_loss": 2.6302945613861084, + "innovation_rms": 0.00396040084454995, + "instruction_rms": 0.003959842161033691, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.412534713745117, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.777933120727539, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16147667169570923, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49447211623191833, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5740494132041931, + "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": 0.04909081384539604, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08679874241352081, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0595308542251587, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15274228155612946, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19161373376846313, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3922130763530731, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5740333199501038, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1186773777008057, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017399546108288238, + "step": 8, + "teaching_rms": 0.00396040084454995, + "traffic_rms": 0.016937892568830458 + }, + { + "batch_loss": 2.8290765285491943, + "innovation_rms": 0.0038182625190261786, + "instruction_rms": 0.0038171153490095485, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.693007469177246, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.359075546264648, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21561501920223236, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5314651727676392, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6767205595970154, + "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": 0.054478615522384644, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12363100051879883, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0829601287841797, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23429271578788757, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21565121412277222, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3934687376022339, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6768028736114502, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1185444593429565, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017427495628907805, + "step": 9, + "teaching_rms": 0.0038182625190261786, + "traffic_rms": 0.017001719118785064 + }, + { + "batch_loss": 2.5880463123321533, + "innovation_rms": 0.003465394002991935, + "instruction_rms": 0.0034637494553891534, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.955960273742676, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.857332229614258, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2514484226703644, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5659844875335693, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5648140907287598, + "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": 0.06370507925748825, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16772453486919403, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1088234186172485, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2693890333175659, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31732702255249023, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3971223533153534, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.56490159034729, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1187033653259277, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017456793794819323, + "step": 10, + "teaching_rms": 0.003465394002991935, + "traffic_rms": 0.017100911886002147 + }, + { + "batch_loss": 2.9120676517486572, + "innovation_rms": 0.0036607588553651846, + "instruction_rms": 0.00365858557153303, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2183773517608643, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.863235473632812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2725955843925476, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5495079755783081, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.676791250705719, + "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": 0.07173214852809906, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19598811864852905, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1426353454589844, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34040260314941406, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30614182353019714, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41327667236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.676742672920227, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1193194389343262, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017553949981062557, + "step": 11, + "teaching_rms": 0.0036607588553651846, + "traffic_rms": 0.01716891303179167 + }, + { + "batch_loss": 2.3453121185302734, + "innovation_rms": 0.0031487276781132327, + "instruction_rms": 0.0031457977312754435, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4616174697875977, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.010584831237793, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28871390223503113, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5399792194366455, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5772626399993896, + "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": 0.09050727635622025, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23833264410495758, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1777777671813965, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4145487844944, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35589009523391724, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4160865545272827, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5772119164466858, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1197394132614136, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01754619169022706, + "step": 12, + "teaching_rms": 0.0031487276781132327, + "traffic_rms": 0.01725171686058308 + }, + { + "batch_loss": 2.6994190216064453, + "innovation_rms": 0.003373649391648118, + "instruction_rms": 0.003369478149029944, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.689697265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.679888725280762, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3185480535030365, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5445080995559692, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3876524269580841, + "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": 0.11326465010643005, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2534089982509613, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2109296321868896, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5710196495056152, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32558366656303406, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41092371940612793, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38762593269348145, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.120368242263794, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01762779564107013, + "step": 13, + "teaching_rms": 0.003373649391648118, + "traffic_rms": 0.017304476001462153 + }, + { + "batch_loss": 2.7510876655578613, + "innovation_rms": 0.0035042131832257006, + "instruction_rms": 0.003499553774720041, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9151456356048584, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.890708923339844, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.338576078414917, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.56352299451828, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5933526754379272, + "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": 0.13574199378490448, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2693963348865509, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2457003593444824, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7066035866737366, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3536158502101898, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40054139494895935, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5933048725128174, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.121161937713623, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01779136008570709, + "step": 14, + "teaching_rms": 0.0035042131832257006, + "traffic_rms": 0.01744689499295311 + }, + { + "batch_loss": 2.362131357192993, + "innovation_rms": 0.003428071903202647, + "instruction_rms": 0.003421213404185238, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.133459091186523, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.957026481628418, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3563518822193146, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5998504757881165, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5144138336181641, + "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": 0.1598844826221466, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2999083995819092, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3332017660140991, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8750140070915222, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4523089528083801, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40302804112434387, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.514520525932312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1214650869369507, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01788926002728816, + "step": 15, + "teaching_rms": 0.003428071903202647, + "traffic_rms": 0.017555592415671947 + }, + { + "batch_loss": 2.4854414463043213, + "innovation_rms": 0.002997299910800856, + "instruction_rms": 0.002986277837101707, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.340120792388916, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.329877853393555, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34727758169174194, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6233876943588257, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5408962368965149, + "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": 0.16356772184371948, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36136531829833984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4398173093795776, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0661556720733643, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6195228695869446, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4304962754249573, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5411584377288818, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.121760368347168, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018036687403786267, + "step": 16, + "teaching_rms": 0.002997299910800856, + "traffic_rms": 0.01779496627327783 + }, + { + "batch_loss": 2.316169500350952, + "innovation_rms": 0.0027712094047524106, + "instruction_rms": 0.002756344328457802, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.538162708282471, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.806221961975098, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3098599314689636, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6288750767707825, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9804195165634155, + "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": 0.20096974074840546, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42905837297439575, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5686290264129639, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2881168127059937, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5415375232696533, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41915008425712585, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.98068767786026, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1220171451568604, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01815356717850905, + "step": 17, + "teaching_rms": 0.0027712094047524106, + "traffic_rms": 0.017945092350512004 + }, + { + "batch_loss": 2.9703874588012695, + "innovation_rms": 0.0030255529859929894, + "instruction_rms": 0.0030066646209446664, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.705987930297852, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.29150676727295, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28353333473205566, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6261765360832214, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4256351590156555, + "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": 0.25246384739875793, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5149409770965576, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7300933599472046, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.614642858505249, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39756229519844055, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4259122908115387, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42536163330078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1226441860198975, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018301077952199755, + "step": 18, + "teaching_rms": 0.0030255529859929894, + "traffic_rms": 0.018075348301460112 + }, + { + "batch_loss": 2.487698554992676, + "innovation_rms": 0.0023368636607162753, + "instruction_rms": 0.002308631880528949, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.851974964141846, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.436269760131836, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3093392252922058, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6217160820960999, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.013292670249939, + "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": 0.3136254549026489, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6116161942481995, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9233149290084839, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9322155714035034, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43424901366233826, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4416898190975189, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0130144357681274, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1234077215194702, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01837380050220077, + "step": 19, + "teaching_rms": 0.0023368636607162753, + "traffic_rms": 0.01823287962685979 + }, + { + "batch_loss": 2.470494508743286, + "innovation_rms": 0.002073525322243719, + "instruction_rms": 0.0020300710793142466, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.972151279449463, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.22831153869629, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3393814265727997, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6255771517753601, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4618285298347473, + "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": 0.38595184683799744, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7232638597488403, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1532094478607178, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2989439964294434, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33937206864356995, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4454145133495331, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4615459144115448, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1240805387496948, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0184307278053481, + "step": 20, + "teaching_rms": 0.002073525322243719, + "traffic_rms": 0.018319235041410076 + }, + { + "batch_loss": 2.733273506164551, + "innovation_rms": 0.0018010392054804423, + "instruction_rms": 0.001733342029915508, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.073139190673828, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.90327262878418, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4404592514038086, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6437256336212158, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5012962222099304, + "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": 0.471828430891037, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.858765721321106, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4318904876708984, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.786811351776123, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4404497742652893, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44381430745124817, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5009135007858276, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1244429349899292, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018496370642630572, + "step": 21, + "teaching_rms": 0.0018010392054804423, + "traffic_rms": 0.0184185372191045 + }, + { + "batch_loss": 2.500389337539673, + "innovation_rms": 0.0016940641201454198, + "instruction_rms": 0.0016056380560251838, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.1829681396484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.64744186401367, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43233516812324524, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6669126749038696, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5367251634597778, + "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": 0.5729570388793945, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0112862586975098, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.764503240585327, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.326127767562866, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.432325541973114, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4344443380832672, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5365577340126038, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1247401237487793, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018506412523495233, + "step": 22, + "teaching_rms": 0.0016940641201454198, + "traffic_rms": 0.018449094439433184 + }, + { + "batch_loss": 2.3803293704986572, + "innovation_rms": 0.001646336957076716, + "instruction_rms": 0.0015299090436573784, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.234825611114502, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.1646842956543, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7957345247268677, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6758971214294434, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5854107141494751, + "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": 0.6922302842140198, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.192732572555542, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.161294460296631, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.967912435531616, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7957727909088135, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42957180738449097, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.585241436958313, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.125049352645874, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01868758778285971, + "step": 23, + "teaching_rms": 0.001646336957076716, + "traffic_rms": 0.01863528068906975 + }, + { + "batch_loss": 2.52144718170166, + "innovation_rms": 0.001625199495385621, + "instruction_rms": 0.0014769351997635995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.318943500518799, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.569000244140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9625133872032166, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6748326420783997, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45750224590301514, + "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": 0.8286917805671692, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3646149635314941, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.605144500732422, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.438500881195068, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9623314142227173, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.43655356764793396, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4573313295841217, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1253998279571533, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01869333110907357, + "step": 24, + "teaching_rms": 0.001625199495385621, + "traffic_rms": 0.01863987528229949 + }, + { + "batch_loss": 2.239332437515259, + "innovation_rms": 0.0015894937165528544, + "instruction_rms": 0.001404602282264127, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.362565994262695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.103736877441406, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7426331639289856, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.672312319278717, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3846636414527893, + "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": 0.982687771320343, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5399596691131592, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.089029312133789, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.838845729827881, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.742688000202179, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4430740177631378, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3844306468963623, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1256226301193237, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018772127645536266, + "step": 25, + "teaching_rms": 0.0015894937165528544, + "traffic_rms": 0.018723313566903564 + }, + { + "batch_loss": 1.9580518007278442, + "innovation_rms": 0.0016153036225517445, + "instruction_rms": 0.0012897660717423097, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.383224010467529, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.53386306762695, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9711865186691284, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6724427342414856, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3775387704372406, + "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": 1.1684763431549072, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8578855991363525, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.697784423828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.087552070617676, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9711176156997681, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4503372609615326, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37730395793914795, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.125893473625183, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018871788189561874, + "step": 26, + "teaching_rms": 0.0016153036225517445, + "traffic_rms": 0.018824448175646213 + }, + { + "batch_loss": 2.6123604774475098, + "innovation_rms": 0.0018034296015150358, + "instruction_rms": 0.0013705869853222875, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.412875175476074, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.955387115478516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2043390274047852, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6741514205932617, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28594231605529785, + "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": 1.3930637836456299, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2458744049072266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.472269535064697, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.744851112365723, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2042696475982666, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45550790429115295, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2858338952064514, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1262096166610718, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018969618444015634, + "step": 27, + "teaching_rms": 0.0018034296015150358, + "traffic_rms": 0.018922128000408003 + }, + { + "batch_loss": 2.4354970455169678, + "innovation_rms": 0.0019298943947879893, + "instruction_rms": 0.00141498653255717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4517502784729, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.890602111816406, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.578152894973755, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.677381157875061, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4210297763347626, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9776297807693481, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6596636772155762, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.665999412536621, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.401047706604004, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.28778076171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5780677795410156, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4658394157886505, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4210452139377594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1264970302581787, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01902201613912799, + "step": 28, + "teaching_rms": 0.0019298943947879893, + "traffic_rms": 0.01897020335122502 + }, + { + "batch_loss": 2.0548391342163086, + "innovation_rms": 0.0016942790899770197, + "instruction_rms": 0.0013183830105852968, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.485004425048828, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 60.65036392211914, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9016048908233643, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6829521059989929, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3598378300666809, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1517928838729858, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.926425814628601, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6676218509674072, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.271411418914795, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.703635215759277, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9015192985534668, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4746878445148468, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3599170446395874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1270047426223755, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01913026291351594, + "step": 29, + "teaching_rms": 0.0016942790899770197, + "traffic_rms": 0.019080256007082956 + }, + { + "batch_loss": 2.14921236038208, + "innovation_rms": 0.0018773510616182881, + "instruction_rms": 0.0013212230683166124, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.504726886749268, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.54925537109375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7411680221557617, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7023407220840454, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32921773195266724, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3111276626586914, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.186889171600342, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6339268684387207, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.073087692260742, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.139284133911133, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7413277626037598, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4830658733844757, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32929736375808716, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2521508932113647, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019353253277135687, + "step": 30, + "teaching_rms": 0.0018773510616182881, + "traffic_rms": 0.01931390070776423 + }, + { + "batch_loss": 2.2747678756713867, + "innovation_rms": 0.001618262545882351, + "instruction_rms": 0.0011595084591987405, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.5228681564331055, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 178.18496704101562, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.582756757736206, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7032449841499329, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23536477982997894, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4609636068344116, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4418561458587646, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.795203924179077, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.815679550170898, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.100984573364258, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.582873821258545, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4820106029510498, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23534038662910461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4019925594329834, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01954003674862032, + "step": 31, + "teaching_rms": 0.001618262545882351, + "traffic_rms": 0.019509025840840542 + }, + { + "batch_loss": 2.2195544242858887, + "innovation_rms": 0.0016007773995341229, + "instruction_rms": 0.001102379063067684, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.528928756713867, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 304.90997314453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.391808032989502, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6826720833778381, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35530704259872437, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.600144386291504, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.692608594894409, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9010825157165527, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.504015922546387, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.859722137451172, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3918652534484863, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4788484573364258, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35529130697250366, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5411790609359741, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01974277606446548, + "step": 32, + "teaching_rms": 0.0016007773995341229, + "traffic_rms": 0.01971175445784028 + }, + { + "batch_loss": 2.180105447769165, + "innovation_rms": 0.0016860711526962841, + "instruction_rms": 0.0011267328473331998, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.9565749168396, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 454.60113525390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5678112506866455, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6617733836174011, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3522018790245056, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7317593097686768, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.939922571182251, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.076472282409668, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.13949203491211, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.7605061531066895, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.567786455154419, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4761657118797302, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3524203896522522, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6727997064590454, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020008042595115228, + "step": 33, + "teaching_rms": 0.0016860711526962841, + "traffic_rms": 0.019975457833722843 + }, + { + "batch_loss": 2.1500606536865234, + "innovation_rms": 0.002051704152640611, + "instruction_rms": 0.0011802514961541657, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.926954746246338, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 579.3190307617188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.712545871734619, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6445955634117126, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23721513152122498, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8600189685821533, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1906368732452393, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5206379890441895, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.745928764343262, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.707279205322266, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7125210762023926, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47709837555885315, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23710328340530396, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.80106520652771, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02008566172061913, + "step": 34, + "teaching_rms": 0.002051704152640611, + "traffic_rms": 0.020054207407489373 + }, + { + "batch_loss": 2.143704891204834, + "innovation_rms": 0.0026700979468299605, + "instruction_rms": 0.0011464707718575283, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.967533588409424, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 708.4779052734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9628026485443115, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6353724002838135, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2793704569339752, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9820369482040405, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4442243576049805, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.420267105102539, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.313044548034668, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.948038101196289, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.962899923324585, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4826149642467499, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2792583107948303, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9230889081954956, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020341709559071584, + "step": 35, + "teaching_rms": 0.0026700979468299605, + "traffic_rms": 0.02030909694304502 + }, + { + "batch_loss": 2.247246026992798, + "innovation_rms": 0.0029471609147755707, + "instruction_rms": 0.0012098884970830799, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.451327323913574, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 802.4945678710938, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8685450553894043, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6297924518585205, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3835262358188629, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1107213497161865, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.829108238220215, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.322413444519043, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.972506523132324, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.909723281860352, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.868664026260376, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49598509073257446, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38341379165649414, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.05177903175354, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020601962095931334, + "step": 36, + "teaching_rms": 0.0029471609147755707, + "traffic_rms": 0.020569471019416915 + }, + { + "batch_loss": 2.1764230728149414, + "innovation_rms": 0.0034977953719120677, + "instruction_rms": 0.0011836753887871615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.938350677490234, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 873.9500732421875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0803699493408203, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6245783567428589, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34303414821624756, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.238877534866333, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.4795708656311035, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.504627704620361, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.914531707763672, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.42025375366211, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.080399990081787, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.515546441078186, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34334897994995117, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.179940938949585, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021141654987636437, + "step": 37, + "teaching_rms": 0.0034977953719120677, + "traffic_rms": 0.021108377152611213 + }, + { + "batch_loss": 2.069282054901123, + "innovation_rms": 0.0041196720934566, + "instruction_rms": 0.0011860226294260806, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.774346351623535, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 927.7402954101562, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5683441162109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6201712489128113, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3521559238433838, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3545844554901123, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.262497901916504, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.829268932342529, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.35669708251953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.421655654907227, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.568673133850098, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5313892364501953, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3521718978881836, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2956535816192627, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021940377122979048, + "step": 38, + "teaching_rms": 0.0041196720934566, + "traffic_rms": 0.021910503718191497 + }, + { + "batch_loss": 2.023991107940674, + "innovation_rms": 0.004810150961925826, + "instruction_rms": 0.0012890781844168452, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.19085121154785, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 917.9185791015625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.307345867156982, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6151611804962158, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29484856128692627, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4168856143951416, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.182026386260986, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.195283889770508, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.24799346923828, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.912961959838867, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.307465553283691, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5390164852142334, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2948645353317261, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3534481525421143, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0225464026711002, + "step": 39, + "teaching_rms": 0.004810150961925826, + "traffic_rms": 0.022511704964267257 + }, + { + "batch_loss": 2.0881340503692627, + "innovation_rms": 0.006267456213965827, + "instruction_rms": 0.0014566753403430271, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.988298416137695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 879.109375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.797306060791016, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.606485903263092, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26181232929229736, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.630608558654785, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.292608737945557, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.10582160949707, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.902379989624023, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.543861389160156, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.797280311584473, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5356004238128662, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26190805435180664, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.599700450897217, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.024006270923752848, + "step": 40, + "teaching_rms": 0.006267456213965827, + "traffic_rms": 0.023962645221908302 + }, + { + "batch_loss": 2.1646947860717773, + "innovation_rms": 0.006216662943098509, + "instruction_rms": 0.0014772431816026691, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.74749755859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 871.1500854492188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.420781135559082, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5958781838417053, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23721927404403687, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.353562116622925, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.595980644226074, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.033721923828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.198854446411133, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.964744567871094, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.42094612121582, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5309981107711792, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2369718849658966, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3226571083068848, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02487412092274266, + "step": 41, + "teaching_rms": 0.006216662943098509, + "traffic_rms": 0.02483410445658554 + }, + { + "batch_loss": 2.2888057231903076, + "innovation_rms": 0.007808894952399441, + "instruction_rms": 0.001454169411034615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.745529174804688, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 894.1853637695312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.559164047241211, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5845506191253662, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2679749131202698, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.252137184143066, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.174574851989746, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.785942077636719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.6347541809082, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.358978271484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.559139251708984, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5302931666374207, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26778504252433777, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.221235275268555, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.027346987236040534, + "step": 42, + "teaching_rms": 0.007808894952399441, + "traffic_rms": 0.02731122352234818 + }, + { + "batch_loss": 1.980799913406372, + "innovation_rms": 0.006970349525408, + "instruction_rms": 0.001339488590129972, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.417829513549805, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 921.02099609375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.261053085327148, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5747223496437073, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4781606197357178, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.1782426834106445, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.911757469177246, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.371829986572266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.2985954284668, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.63841247558594, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.261083602905273, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5340538024902344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47797054052352905, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.147343635559082, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.027919779052614377, + "step": 43, + "teaching_rms": 0.006970349525408, + "traffic_rms": 0.02788715527138692 + }, + { + "batch_loss": 2.094031810760498, + "innovation_rms": 0.008484063545845608, + "instruction_rms": 0.00126379996194369, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.23892593383789, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1140.7022705078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.844964981079102, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.570979654788971, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33251920342445374, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.273938179016113, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.858979225158691, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.47222137451172, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.3895149230957, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60.9091911315918, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.844879150390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5378659963607788, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3323289155960083, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.243042469024658, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03066690041684519, + "step": 44, + "teaching_rms": 0.008484063545845608, + "traffic_rms": 0.030637448236769577 + }, + { + "batch_loss": 2.113565444946289, + "innovation_rms": 0.01389375149227664, + "instruction_rms": 0.0013053018963318334, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.41667556762695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1311.717529296875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.29023551940918, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5731181502342224, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34341105818748474, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.033836364746094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.32360076904297, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.646215438842773, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.0091438293457, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86.19630432128906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.290149688720703, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5467309355735779, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34319841861724854, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.002943992614746, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03697363718595933, + "step": 45, + "teaching_rms": 0.01389375149227664, + "traffic_rms": 0.036951972847095185 + }, + { + "batch_loss": 2.218254804611206, + "innovation_rms": 0.012752124185726351, + "instruction_rms": 0.0013223192030646467, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.95226287841797, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2009.67431640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.230445861816406, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5748292803764343, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25639164447784424, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.696212768554688, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.09649658203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.728960037231445, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.46430587768555, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94.5516357421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.230361938476562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5516732335090637, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2561788260936737, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.665323257446289, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04324081058537506, + "step": 46, + "teaching_rms": 0.012752124185726351, + "traffic_rms": 0.04322166669805092 + }, + { + "batch_loss": 2.2012522220611572, + "innovation_rms": 0.014034250778132348, + "instruction_rms": 0.0011102263265685376, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.767120361328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3639.16943359375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.621585845947266, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5815579891204834, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2537829875946045, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.35837173461914, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.13141441345215, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.34916877746582, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.35161590576172, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.87312316894531, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.62161636352539, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5547119975090027, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25375601649284363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.327485084533691, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.050198005699634514, + "step": 47, + "teaching_rms": 0.014034250778132348, + "traffic_rms": 0.05018659519157906 + }, + { + "batch_loss": 2.149991989135742, + "innovation_rms": 0.019163230673969335, + "instruction_rms": 0.0013699885057431548, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.45481491088867, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5168.36279296875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.910688400268555, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5863428711891174, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24006980657577515, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.127617835998535, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.676362991333008, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.449493408203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.00694274902344, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.55328369140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.910663604736328, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5499742031097412, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2400428205728531, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.096734046936035, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06070775015290519, + "step": 48, + "teaching_rms": 0.019163230673969335, + "traffic_rms": 0.060690864324291625 + }, + { + "batch_loss": 2.30002760887146, + "innovation_rms": 0.02962055770264825, + "instruction_rms": 0.0015620048400142915, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.815826416015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5883.9912109375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.57993698120117, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5800383687019348, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27494296431541443, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.958843231201172, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.305944442749023, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.295814514160156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.86210632324219, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178.55162048339844, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.5799560546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5413018465042114, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27476924657821655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.927962303161621, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06996748747203076, + "step": 49, + "teaching_rms": 0.02962055770264825, + "traffic_rms": 0.06995039323057524 + }, + { + "batch_loss": 2.076509475708008, + "innovation_rms": 0.0314236575259601, + "instruction_rms": 0.001326260176913414, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.24105834960938, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8316.8525390625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.66985321044922, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5755254626274109, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3433606028556824, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.08775520324707, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.820045471191406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.141014099121094, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.30555725097656, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.43450927734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.6698112487793, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5462146401405334, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34318679571151733, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.074254989624023, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.08389990453641505, + "step": 50, + "teaching_rms": 0.0314236575259601, + "traffic_rms": 0.08388886030241736 + }, + { + "batch_loss": 2.223979949951172, + "innovation_rms": 0.03917195020240213, + "instruction_rms": 0.0014184019020775972, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.89340209960938, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10280.640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.032958984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5739208459854126, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3775888681411743, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.639802932739258, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.654998779296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.34954071044922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.32089233398438, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280.1533203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.032958984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5679501891136169, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3776935636997223, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.68162727355957, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.10728938546119689, + "step": 51, + "teaching_rms": 0.03917195020240213, + "traffic_rms": 0.10727992348893481 + }, + { + "batch_loss": 2.2050654888153076, + "innovation_rms": 0.03475437566901251, + "instruction_rms": 0.001282495599531992, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 106.76818084716797, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14561.1044921875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.03104782104492, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5847790837287903, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4720548987388611, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.64667510986328, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.20712661743164, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.5212631225586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 180.07196044921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287.5106201171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.03093719482422, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5538175106048584, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47215962409973145, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.688495635986328, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.13245181588341248, + "step": 52, + "teaching_rms": 0.03475437566901251, + "traffic_rms": 0.13244551303172597 + }, + { + "batch_loss": 2.3213300704956055, + "innovation_rms": 0.03491952852438637, + "instruction_rms": 0.0014226429436582219, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 97.63139343261719, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22273.15234375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.076934814453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5831199884414673, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3543420433998108, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.80696487426758, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.38441848754883, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.77291870117188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.89044189453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288.1847839355469, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.07693862915039, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5383736491203308, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35451602935791016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.84878158569336, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.1476267888860882, + "step": 53, + "teaching_rms": 0.03491952852438637, + "traffic_rms": 0.14762094256944758 + }, + { + "batch_loss": 2.2369532585144043, + "innovation_rms": 0.039627361572437826, + "instruction_rms": 0.0014873745161408756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.84725952148438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34790.8984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.20163345336914, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5990620851516724, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36908483505249023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.03059387207031, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.32781982421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.43402862548828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.25772094726562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 293.6727600097656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.201637268066406, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.538883626461029, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3690233826637268, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.07240676879883, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.1761369815131369, + "step": 54, + "teaching_rms": 0.039627361572437826, + "traffic_rms": 0.1761316552005646 + }, + { + "batch_loss": 2.261955976486206, + "innovation_rms": 0.04968965793398473, + "instruction_rms": 0.001424261045861715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.66252899169922, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 47201.36328125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.58882904052734, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6053868532180786, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44317594170570374, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.53961944580078, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.41163635253906, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.83814239501953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 269.7548828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 314.97174072265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.5890884399414, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5397056937217712, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44268956780433655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.58142852783203, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2098926420255104, + "step": 55, + "teaching_rms": 0.04968965793398473, + "traffic_rms": 0.20988795976520372 + }, + { + "batch_loss": 2.186253547668457, + "innovation_rms": 0.06537745157062222, + "instruction_rms": 0.0013783486608361392, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.6649169921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 55581.7421875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.67977905273438, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5937149524688721, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3948226571083069, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.13407516479492, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.61727142333984, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.05636596679688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 307.7771301269531, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 380.22247314453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.68003845214844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5447754859924316, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3948236107826233, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.175880432128906, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24748126488890285, + "step": 56, + "teaching_rms": 0.06537745157062222, + "traffic_rms": 0.2474775949779154 + }, + { + "batch_loss": 2.171172857284546, + "innovation_rms": 0.0777095716770648, + "instruction_rms": 0.0014854496711326972, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 148.51785278320312, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68068.8671875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.44806671142578, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5856667757034302, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3823796808719635, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.0644645690918, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.25909423828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.4182586669922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 352.72503662109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 449.47918701171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.44831848144531, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5496749877929688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3823806047439575, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.106266021728516, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2824444708560555, + "step": 57, + "teaching_rms": 0.0777095716770648, + "traffic_rms": 0.28244019406643434 + }, + { + "batch_loss": 2.178112030029297, + "innovation_rms": 0.09552952453079802, + "instruction_rms": 0.001264784197573079, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.36729431152344, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80533.2734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.60992431640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5774238705635071, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40072184801101685, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.06906127929688, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.86363220214844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.04541015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 407.1783142089844, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 544.53271484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.60992431640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5486575961112976, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.400913268327713, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.11085510253906, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3440039649615171, + "step": 58, + "teaching_rms": 0.09552952453079802, + "traffic_rms": 0.344001418387197 + }, + { + "batch_loss": 2.350351572036743, + "innovation_rms": 0.13753439533716155, + "instruction_rms": 0.0011965637315899036, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 251.0474853515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 92872.9453125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.20889282226562, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5750585794448853, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.442069411277771, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.330078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 143.2645263671875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.00889587402344, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 475.4728088378906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 682.9448852539062, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.20909118652344, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5534111857414246, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44226086139678955, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.37186431884766, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.4417778386130405, + "step": 59, + "teaching_rms": 0.13753439533716155, + "traffic_rms": 0.4417761925420376 + }, + { + "batch_loss": 2.1470208168029785, + "innovation_rms": 0.1224258332950325, + "instruction_rms": 0.0012505690829827556, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 333.063720703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 114764.6484375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.8907470703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5852004885673523, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42212986946105957, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.10372161865234, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.29859924316406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 250.34072875976562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 555.0111083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 795.3831176757812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.89077758789062, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5655608177185059, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4219185411930084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.1007308959961, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.5481212397624705, + "step": 60, + "teaching_rms": 0.1224258332950325, + "traffic_rms": 0.5481196652883795 + }, + { + "batch_loss": 2.330536365509033, + "innovation_rms": 0.09319503908578736, + "instruction_rms": 0.00147578211005842, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.84869384765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 224914.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.55442810058594, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5993178486824036, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.381372332572937, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.08106994628906, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.80856323242188, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 265.099609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 633.98486328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 789.7377319335938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151.55435180664062, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5808209776878357, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3811803460121155, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.07807922363281, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.6706132240434595, + "step": 61, + "teaching_rms": 0.09319503908578736, + "traffic_rms": 0.670611611956037 + }, + { + "batch_loss": 2.179542303085327, + "innovation_rms": 0.08607286931745027, + "instruction_rms": 0.0012054418830798506, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 371.1307373046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 483986.90625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.03321838378906, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6092509627342224, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3395182490348816, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.06111907958984, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 222.26278686523438, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 274.5422058105469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 710.2764892578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 762.9163818359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.0332489013672, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5905415415763855, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3394213914871216, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.02616882324219, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7770570923177574, + "step": 62, + "teaching_rms": 0.08607286931745027, + "traffic_rms": 0.7770561194123637 + }, + { + "batch_loss": 2.355299711227417, + "innovation_rms": 0.1219502363820227, + "instruction_rms": 0.0013135121920326254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 414.6972351074219, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 732337.875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.14817810058594, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6120560169219971, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3999806344509125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.5018310546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 252.62091064453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303.58123779296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 791.3126831054688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 810.3621215820312, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.1481475830078, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.600408136844635, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39988377690315247, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.46366119384766, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.8784924220027918, + "step": 63, + "teaching_rms": 0.1219502363820227, + "traffic_rms": 0.8784915185673015 + }, + { + "batch_loss": 2.4577088356018066, + "innovation_rms": 0.18292250063389637, + "instruction_rms": 0.0015253080731082679, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 521.8568115234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 889094.8125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.4690246582031, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6056860089302063, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35385531187057495, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.60222625732422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 289.1160583496094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 364.95135498046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 892.3173828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010.0470581054688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.468994140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5974184274673462, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35378625988960266, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.5640640258789, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.9666898539232944, + "step": 64, + "teaching_rms": 0.18292250063389637, + "traffic_rms": 0.9666886332423121 + }, + { + "batch_loss": 2.1502790451049805, + "innovation_rms": 0.2395597752142347, + "instruction_rms": 0.0016061524795083738, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 643.88623046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1007862.5625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339.13330078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5977352857589722, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34536877274513245, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.37168884277344, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 334.02880859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 449.1273498535156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1026.2135009765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1338.960693359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339.13311767578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5932496190071106, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34529972076416016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.17832946777344, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.0662173283181016, + "step": 65, + "teaching_rms": 0.2395597752142347, + "traffic_rms": 1.0662160125986098 + }, + { + "batch_loss": 2.1653285026550293, + "innovation_rms": 0.30185842804155727, + "instruction_rms": 0.0016440217727242642, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 699.1270751953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1101363.375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 408.7005310058594, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5911673903465271, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30052807927131653, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.4813995361328, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 389.96826171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 559.3944091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1208.8536376953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1826.4014892578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 408.7008056640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5909680724143982, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30045902729034424, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.6754913330078, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.2057976545558444, + "step": 66, + "teaching_rms": 0.30185842804155727, + "traffic_rms": 1.2057964893788804 + }, + { + "batch_loss": 2.385305404663086, + "innovation_rms": 0.39513282847773284, + "instruction_rms": 0.0017128865738780495, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 671.4766235351562, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1188587.75, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 504.4005126953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5922030210494995, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2986127734184265, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.7590789794922, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 460.039306640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 700.7105102539062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1461.83642578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2529.827392578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 504.40057373046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5947690010070801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2986266016960144, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.95315551757812, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.3373792742912878, + "step": 67, + "teaching_rms": 0.39513282847773284, + "traffic_rms": 1.3373781858374005 + }, + { + "batch_loss": 2.2395174503326416, + "innovation_rms": 0.5836487281604986, + "instruction_rms": 0.0014054815394890824, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 752.0988159179688, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1270168.375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1192.42626953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5966671109199524, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28657981753349304, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 277.9002685546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 551.9197998046875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 918.8051147460938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.0164794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4021.800537109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1192.4261474609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.601499617099762, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28659361600875854, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 278.0943298339844, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.5028999256097513, + "step": 68, + "teaching_rms": 0.5836487281604986, + "traffic_rms": 1.5028992448797995 + }, + { + "batch_loss": 2.14119291305542, + "innovation_rms": 0.6495890141946032, + "instruction_rms": 0.0017182062320251778, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 669.3395385742188, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1488159.75, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1614.90283203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5868902206420898, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24361328780651093, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 356.85919189453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 659.0432739257812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1071.234619140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2352.853759765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4888.373046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1614.9029541015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6006585359573364, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2437165528535843, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 356.83477783203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1.829555848720569, + "step": 69, + "teaching_rms": 0.6495890141946032, + "traffic_rms": 1.829555045846844 + }, + { + "batch_loss": 2.3493523597717285, + "innovation_rms": 0.6269034234799594, + "instruction_rms": 0.0015711001557352092, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 902.7325439453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2632758.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1346.2122802734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5908154249191284, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3893306255340576, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 491.48040771484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 783.0775756835938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1240.3431396484375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2910.608642578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5577.54931640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1346.21240234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6121474504470825, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3895353674888611, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 491.4560241699219, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.235180989729384, + "step": 70, + "teaching_rms": 0.6269034234799594, + "traffic_rms": 2.235180347580299 + }, + { + "batch_loss": 2.261275291442871, + "innovation_rms": 0.8695531387366199, + "instruction_rms": 0.001208859965577524, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1464.1781005859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4387562.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1726.467529296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6029821634292603, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.377994179725647, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 649.4070434570312, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 933.6734008789062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1505.958251953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3620.5302734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7099.216796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1726.46728515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6335719227790833, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37774747610092163, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 649.3826904296875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.7506000575641743, + "step": 71, + "teaching_rms": 0.8695531387366199, + "traffic_rms": 2.7505997849321324 + }, + { + "batch_loss": 2.1747841835021973, + "innovation_rms": 1.1588271976251836, + "instruction_rms": 0.0012674669368768206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.29150390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6220721.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4078.958984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.613629937171936, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3211609721183777, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 841.6674194335938, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1120.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.3284912109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4555.3076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9347.775390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4078.958984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.655195415019989, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32091423869132996, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 841.64306640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.263062623259847, + "step": 72, + "teaching_rms": 1.1588271976251836, + "traffic_rms": 3.2630623801870806 + }, + { + "batch_loss": 2.151531457901001, + "innovation_rms": 0.7987336158584659, + "instruction_rms": 0.0013619391597648482, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1258.6680908203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17237382.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3012.08154296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6178877949714661, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3247915506362915, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1070.98193359375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1319.302001953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1983.957275390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5474.0927734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9187.8525390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3012.08154296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6678414344787598, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32454484701156616, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1071.1759033203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3.887958264298071, + "step": 73, + "teaching_rms": 0.7987336158584659, + "traffic_rms": 3.8879580232431827 + }, + { + "batch_loss": 2.260737657546997, + "innovation_rms": 0.9226493277403662, + "instruction_rms": 0.0015736899532936784, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1468.5057373046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 45417976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2417.47021484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6137014627456665, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.464706689119339, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1287.4324951171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1528.80908203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2095.0712890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6365.5849609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8914.9208984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2417.470458984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6678386926651001, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46454155445098877, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1287.62646484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.430527235415793, + "step": 74, + "teaching_rms": 0.9226493277403662, + "traffic_rms": 4.4305269753560665 + }, + { + "batch_loss": 2.2842791080474854, + "innovation_rms": 1.0550407383555969, + "instruction_rms": 0.0013209204915249476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2195.232666015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 101667912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2499.24267578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6253195405006409, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6098248362541199, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1484.1405029296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1748.3876953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2195.785888671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7220.00439453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8544.1953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2499.24267578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6718381643295288, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6097379326820374, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1484.33447265625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.057901251038979, + "step": 75, + "teaching_rms": 1.0550407383555969, + "traffic_rms": 5.057901062869375 + }, + { + "batch_loss": 2.3051445484161377, + "innovation_rms": 1.5675010325027794, + "instruction_rms": 0.0012605197744965453, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4029.515869140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 149303680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2017.783203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6378977298736572, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5458269715309143, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1665.5450439453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1991.174072265625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2427.8642578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8087.1181640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8671.138671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2017.783447265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6738296747207642, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5457401275634766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1665.739013671875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5.73901187567119, + "step": 76, + "teaching_rms": 1.5675010325027794, + "traffic_rms": 5.739011736846213 + }, + { + "batch_loss": 2.1889281272888184, + "innovation_rms": 1.8345610277668998, + "instruction_rms": 0.0012864786984137725, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5553.09423828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 166578064.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2119.27001953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6490288972854614, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7354893684387207, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833.177734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2284.217041015625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2930.4287109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9094.6787109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10075.6064453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2119.2705078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6739583611488342, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7355031967163086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833.3717041015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.860218613244907, + "step": 77, + "teaching_rms": 1.8345610277668998, + "traffic_rms": 6.86021848707633 + }, + { + "batch_loss": 2.046233892440796, + "innovation_rms": 2.4133185794522385, + "instruction_rms": 0.0017912308444162613, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6450.61376953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 187024112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2559.340576171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.658513069152832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9084929823875427, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2042.9111328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2651.092041015625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3668.74951171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10405.1279296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13104.4921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2559.340576171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6719747185707092, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9085068106651306, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2043.1051025390625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8.210480116353422, + "step": 78, + "teaching_rms": 2.4133185794522385, + "traffic_rms": 8.210479916761473 + }, + { + "batch_loss": 2.1365206241607666, + "innovation_rms": 3.759841121373812, + "instruction_rms": 0.002066350363665721, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5830.17578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 207946672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5704.6953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6619586944580078, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34567052125930786, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2442.21826171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3136.153076171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4850.609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12440.9638671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20358.359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5704.6953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6733708381652832, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3457801043987274, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2442.412353515625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9.770264868498911, + "step": 79, + "teaching_rms": 3.759841121373812, + "traffic_rms": 9.770264651270093 + }, + { + "batch_loss": 2.1675233840942383, + "innovation_rms": 4.1055299259256195, + "instruction_rms": 0.001512933769594625, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3981.98291015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 246042672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7979.45849609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6593950390815735, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1347416639328003, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2840.773681640625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3756.745361328125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6205.923828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15190.3427734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27493.7890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7979.458984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6816427111625671, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1347278356552124, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2840.9677734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11.308089317608141, + "step": 80, + "teaching_rms": 4.1055299259256195, + "traffic_rms": 11.308089215218878 + }, + { + "batch_loss": 2.1546154022216797, + "innovation_rms": 4.715095984035374, + "instruction_rms": 0.0013237305324553601, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4937.669921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 318386144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10954.2939453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6844499111175537, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.046999454498291, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3348.247802734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4509.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7524.42236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18567.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33772.19140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10954.2939453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6875607371330261, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0469856262207031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3348.44189453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13.428791351400783, + "step": 81, + "teaching_rms": 4.715095984035374, + "traffic_rms": 13.428791287616692 + }, + { + "batch_loss": 2.1504733562469482, + "innovation_rms": 6.254674566203554, + "instruction_rms": 0.0012856370069966219, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6880.4052734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 449900160.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17475.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7135693430900574, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7049321532249451, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4843.06884765625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5417.5302734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9083.4296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22791.9765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42244.140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17475.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7002299427986145, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7049183249473572, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4842.94384765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16.410027658417604, + "step": 82, + "teaching_rms": 6.254674566203554, + "traffic_rms": 16.410027604629512 + }, + { + "batch_loss": 2.213594913482666, + "innovation_rms": 10.636792141766012, + "instruction_rms": 0.001301583949711144, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8788.2138671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 495442912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30059.080078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7431167960166931, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6550516486167908, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7848.97705078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6679.892578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12623.62109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 30248.458984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74564.828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30059.080078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7106144428253174, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6551101207733154, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7848.85205078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 21.204379168828304, + "step": 83, + "teaching_rms": 10.636792141766012, + "traffic_rms": 21.204379130454058 + }, + { + "batch_loss": 2.032874584197998, + "innovation_rms": 9.159578741384074, + "instruction_rms": 0.0012096891267599664, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9802.12890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1385715072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35718.2421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.755475640296936, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5382646322250366, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11420.8017578125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8092.57421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14126.814453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 37963.234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77147.7421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35718.2421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7189054489135742, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5383231043815613, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11420.6767578125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26.111597144332784, + "step": 84, + "teaching_rms": 9.159578741384074, + "traffic_rms": 26.11159711873176 + }, + { + "batch_loss": 2.2230029106140137, + "innovation_rms": 9.520952588962345, + "instruction_rms": 0.001213939530050377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11594.697265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3519141376.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41370.1953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7519635558128357, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.935483992099762, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15557.8212890625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9643.974609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15779.546875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 45782.4609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78192.28125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41370.1953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7226455211639404, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9353963136672974, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15557.6962890625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33.484404869746584, + "step": 85, + "teaching_rms": 9.520952588962345, + "traffic_rms": 33.484404892353204 + }, + { + "batch_loss": 2.1587328910827637, + "innovation_rms": 11.29230903806411, + "instruction_rms": 0.0009940886895193884, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14018.3505859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8833886208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43710.3203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.750001847743988, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1093448400497437, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19928.853515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11309.830078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20737.70703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 53420.23828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76377.7890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43710.3203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7213025689125061, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1092571020126343, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19928.728515625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 40.385243593614334, + "step": 86, + "teaching_rms": 11.29230903806411, + "traffic_rms": 40.38524356577148 + }, + { + "batch_loss": 2.094897985458374, + "innovation_rms": 12.847532182899217, + "instruction_rms": 0.0010423550380816609, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15962.51171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 18729711616.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46293.34765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7620991468429565, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6766266226768494, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 24558.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13119.7490234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26004.1640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 61018.12890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78673.546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46293.34765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7220203280448914, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6765681505203247, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 24558.0625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 54.802154388362965, + "step": 87, + "teaching_rms": 12.847532182899217, + "traffic_rms": 54.802154383442236 + }, + { + "batch_loss": 2.2778687477111816, + "innovation_rms": 24.62097868845895, + "instruction_rms": 0.0012422844335131433, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19795.62109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 21676392448.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82495.5234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7778516411781311, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7167940139770508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32807.73828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16422.21484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37676.38671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 71357.6171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135178.84375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82495.5234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7247835993766785, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7167802453041077, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32807.61328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 70.35819347690519, + "step": 88, + "teaching_rms": 24.62097868845895, + "traffic_rms": 70.35819346611964 + }, + { + "batch_loss": 2.1623926162719727, + "innovation_rms": 22.41003920175176, + "instruction_rms": 0.0010292390642398328, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26618.74609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24879425536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91342.734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8157540559768677, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2867608070373535, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41942.01171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20620.765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41985.515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 84108.328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135253.296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91342.734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7277608513832092, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.286848545074463, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41941.88671875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 92.59608098431708, + "step": 89, + "teaching_rms": 22.41003920175176, + "traffic_rms": 92.59608097814245 + }, + { + "batch_loss": 2.414829730987549, + "innovation_rms": 26.315503350096215, + "instruction_rms": 0.0011915968554310574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33439.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 33025630208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112497.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.81610107421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1185035705566406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 53191.77734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 25337.1640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47163.9765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 99226.71875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 151183.890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112497.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7349613308906555, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1185619831085205, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 53191.65234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 117.71108468968957, + "step": 90, + "teaching_rms": 26.315503350096215, + "traffic_rms": 117.71108468424714 + }, + { + "batch_loss": 2.1660728454589844, + "innovation_rms": 43.30848678228497, + "instruction_rms": 0.0016193961033934944, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 45089.9140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46295224320.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142387.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8357128500938416, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6530739665031433, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67430.484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31280.052734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59428.87890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 123422.34375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 241956.21875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142387.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7430944442749023, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.653238832950592, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67430.359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 142.9570958020584, + "step": 91, + "teaching_rms": 43.30848678228497, + "traffic_rms": 142.9570957927685 + }, + { + "batch_loss": 2.3302981853485107, + "innovation_rms": 61.450299536813866, + "instruction_rms": 0.0019080243925797778, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56138.484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 98963070976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 215152.796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.82448410987854, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6679269075393677, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 88945.765625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 39021.41796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77413.6484375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 162654.78125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 392324.40625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 215152.796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7399786710739136, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6677619218826294, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 88945.640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 158.08538059414556, + "step": 92, + "teaching_rms": 61.450299536813866, + "traffic_rms": 158.08538058318913 + }, + { + "batch_loss": 2.289055109024048, + "innovation_rms": 79.29381961372646, + "instruction_rms": 0.0018615577274334273, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 63982.69140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 231626178560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272476.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.805288553237915, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0708026885986328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116193.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 49966.515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109450.9921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 213050.953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 503961.75, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272476.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7345147728919983, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0708270072937012, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 116193.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 212.96689371766215, + "step": 93, + "teaching_rms": 79.29381961372646, + "traffic_rms": 212.96689371013898 + }, + { + "batch_loss": 2.114544153213501, + "innovation_rms": 115.2150973096717, + "instruction_rms": 0.0015250455103755502, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 82534.203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 421359517696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 347851.15625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8042973279953003, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9497658014297485, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150978.484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 62855.2578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128887.421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 303177.59375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 901266.375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 347851.15625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7311549782752991, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9497075080871582, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150978.359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 287.05325843109296, + "step": 94, + "teaching_rms": 115.2150973096717, + "traffic_rms": 287.0532584270259 + }, + { + "batch_loss": 2.2192189693450928, + "innovation_rms": 163.35327445679317, + "instruction_rms": 0.001513578503857841, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 105702.984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 946337284096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 533381.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8088176846504211, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4551559686660767, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 204316.671875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 78249.3671875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 172335.515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 450037.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1468597.75, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 533381.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.731292724609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4550683498382568, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 204316.546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 369.611217063689, + "step": 95, + "teaching_rms": 163.35327445679317, + "traffic_rms": 369.6112170609297 + }, + { + "batch_loss": 2.6090149879455566, + "innovation_rms": 231.70661439144425, + "instruction_rms": 0.002197867743432637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 131986.15625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1736600322048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 615591.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7994627952575684, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9569692015647888, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 257612.859375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 98773.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 248872.125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 671690.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2216526.25, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 615591.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7384127378463745, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9570273756980896, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 257612.734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 475.6817328571138, + "step": 96, + "teaching_rms": 231.70661439144425, + "traffic_rms": 475.6817328521311 + }, + { + "batch_loss": 2.6097428798675537, + "innovation_rms": 281.67040479270025, + "instruction_rms": 0.002223298032425685, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 171624.671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2946237726720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2261227.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8048313856124878, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.203185558319092, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 483735.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 129400.5234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 336342.34375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 953734.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2820440.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2261227.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7490577101707458, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.203273057937622, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 483735.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 584.2544067058077, + "step": 97, + "teaching_rms": 281.67040479270025, + "traffic_rms": 584.2544067014094 + }, + { + "batch_loss": 2.3778159618377686, + "innovation_rms": 430.313959168619, + "instruction_rms": 0.0015263917283635997, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 202249.109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8631956275200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1770793.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8100844621658325, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4584083557128906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 660815.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 175884.921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 464844.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1361760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4080259.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1770793.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7559970617294312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4583841562271118, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 660814.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 834.5573790601505, + "step": 98, + "teaching_rms": 430.313959168619, + "traffic_rms": 834.5573790589469 + }, + { + "batch_loss": 2.5858073234558105, + "innovation_rms": 584.5332668568916, + "instruction_rms": 0.0014098311529539222, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 296449.1875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12606957944832.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2493635.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8251472115516663, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6139707565307617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 910178.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 243515.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 676302.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1926934.875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5651748.5, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2493635.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7565494775772095, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6139576435089111, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 910178.4375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1138.1307117032663, + "step": 99, + "teaching_rms": 584.5332668568916, + "traffic_rms": 1138.130711702514 + }, + { + "batch_loss": 2.508331060409546, + "innovation_rms": 880.018156659301, + "instruction_rms": 0.0015371814389496437, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 497962.96875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20487186939904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3194838.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.83931964635849, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1851619482040405, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1229662.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 344223.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1007080.5625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2793272.25, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8663374.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3194838.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7561677098274231, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1851751804351807, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1229662.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1521.9120818645938, + "step": 100, + "teaching_rms": 880.018156659301, + "traffic_rms": 1521.9120818639467 + }, + { + "batch_loss": 2.4572360515594482, + "innovation_rms": 1479.5287084197923, + "instruction_rms": 0.0015832683255704637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 663406.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 37392891248640.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3850752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8579145073890686, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4304280281066895, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1614737.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 497441.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1532181.25, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4305567.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15122954.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3850752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7568698525428772, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4304412603378296, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1614737.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2160.5321709007044, + "step": 101, + "teaching_rms": 1479.5287084197923, + "traffic_rms": 2160.5321709002515 + }, + { + "batch_loss": 2.229778289794922, + "innovation_rms": 2322.8870417647363, + "instruction_rms": 0.0017934482059520044, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 920853.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66801434099712.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7966108.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8679431080818176, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6864440441131592, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2411348.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 732909.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2354686.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6841905.5, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25363378.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7966108.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7587044835090637, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.686356544494629, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2411348.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3216.826258824761, + "step": 102, + "teaching_rms": 2322.8870417647363, + "traffic_rms": 3216.826258824257 + }, + { + "batch_loss": 2.2167253494262695, + "innovation_rms": 3414.8721376736744, + "instruction_rms": 0.0022971630218154367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1020832.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 113226486382592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6490861.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8885312676429749, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8534589409828186, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3060434.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1092199.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3592899.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10820879.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39789736.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6490861.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7628491520881653, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8533714413642883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3060434.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4835.040408860783, + "step": 103, + "teaching_rms": 3414.8721376736744, + "traffic_rms": 4835.040408860215 + }, + { + "batch_loss": 2.467944383621216, + "innovation_rms": 5113.775341573861, + "instruction_rms": 0.002489085558181676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1004726.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 202613790343168.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10005496.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.911106526851654, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2061071395874023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4060984.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1642539.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5503399.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16831900.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60110208.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10005496.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7655068039894104, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2060937881469727, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4060984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6932.472372337881, + "step": 104, + "teaching_rms": 5113.775341573861, + "traffic_rms": 6932.4723723374245 + }, + { + "batch_loss": 2.4973130226135254, + "innovation_rms": 8071.36963190938, + "instruction_rms": 0.002689127328550152, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1389598.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 413607766523904.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44700672.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9350730180740356, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.788583755493164, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4900401.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2483199.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8406594.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 26177156.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93452560.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44700672.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7606571912765503, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7885702848434448, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4900401.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10361.937910321914, + "step": 105, + "teaching_rms": 8071.36963190938, + "traffic_rms": 10361.937910321578 + }, + { + "batch_loss": 2.82802152633667, + "innovation_rms": 8680.652420241639, + "instruction_rms": 0.0025559752921769183, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7206579.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 773981493788672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110972448.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9227966666221619, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.186983585357666, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15173382.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3661046.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11778474.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 37432244.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112550896.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110972448.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7657037377357483, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1868882179260254, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15173382.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12446.961335765594, + "step": 106, + "teaching_rms": 8680.652420241639, + "traffic_rms": 12446.961335765363 + }, + { + "batch_loss": 3.2417852878570557, + "innovation_rms": 15857.158174826232, + "instruction_rms": 0.001946092426636638, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5535460.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2709807341502464.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91845480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9311662912368774, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.523839235305786, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 24357930.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5436272.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17752256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 54451080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170188352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91845480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.772284746170044, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.523740530014038, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 24357930.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20131.017610297346, + "step": 107, + "teaching_rms": 15857.158174826232, + "traffic_rms": 20131.017610297265 + }, + { + "batch_loss": 3.462362051010132, + "innovation_rms": 23952.904385214246, + "instruction_rms": 0.0032688808067958333, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7282311.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.572028897296384e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202058080.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.954136848449707, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.673588514328003, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44563740.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8135763.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26994908.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 80619144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 261680640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202058080.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7777705192565918, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6734895706176758, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44563740.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28881.268259420136, + "step": 108, + "teaching_rms": 23952.904385214246, + "traffic_rms": 28881.268259419965 + }, + { + "batch_loss": 4.353660583496094, + "innovation_rms": 39030.36917194883, + "instruction_rms": 0.004845333083191455, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15001050.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.445073076047053e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176917504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.988399863243103, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4568700790405273, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 62255492.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12239813.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41040492.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 123747832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 431286912.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176917504.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7752038836479187, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4569578170776367, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 62255492.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 46324.9226911559, + "step": 109, + "teaching_rms": 39030.36917194883, + "traffic_rms": 46324.922691155654 + }, + { + "batch_loss": 3.9591150283813477, + "innovation_rms": 61120.69790238125, + "instruction_rms": 0.004523991380964864, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 28428742.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1774525554019533e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285049984.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9639426469802856, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3104701042175293, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 70345424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18559010.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 63191960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 193919104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701712704.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285049984.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7737102508544922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.310509204864502, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 70345424.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 71135.973152514, + "step": 110, + "teaching_rms": 61120.69790238125, + "traffic_rms": 71135.97315251385 + }, + { + "batch_loss": 5.738287448883057, + "innovation_rms": 88820.76259041445, + "instruction_rms": 0.005664955751310287, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 44808128.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.649231644763095e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 420077760.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9796484112739563, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.478951930999756, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112353200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 28406720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98477096.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 299147712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1052286080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 420077760.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7751225233078003, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.478938102722168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112353200.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 100324.95577168374, + "step": 111, + "teaching_rms": 88820.76259041445, + "traffic_rms": 100324.9557716836 + }, + { + "batch_loss": 5.317905426025391, + "innovation_rms": 158564.74789335698, + "instruction_rms": 0.00820598932508402, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 64474980.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5324910604569805e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 503331520.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.994400680065155, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.528407096862793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 162686352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 44286632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158799136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 481448320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1823006080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 503331520.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8032103180885315, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.528456211090088, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 162686352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 174438.92979995586, + "step": 112, + "teaching_rms": 158564.74789335698, + "traffic_rms": 174438.9297999557 + }, + { + "batch_loss": 4.886641025543213, + "innovation_rms": 225591.52128785665, + "instruction_rms": 0.01163083098251544, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 212420208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2769410777405194e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1096630784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.009907841682434, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4557337760925293, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 197626720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 68308768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 240221344.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 755753344.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2743050240.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1096630784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8210799098014832, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4557831287384033, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 197626720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 270495.31784648565, + "step": 113, + "teaching_rms": 225591.52128785665, + "traffic_rms": 270495.3178464854 + }, + { + "batch_loss": 8.321587562561035, + "innovation_rms": 385539.8513420409, + "instruction_rms": 0.012199682289914573, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 164245088.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.812814067139084e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2597344000.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0084228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.6964545249938965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 457361120.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 107814880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 395061120.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1207787136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4520337920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2597344000.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8181883096694946, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.696358680725098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 457361120.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 412564.81509303156, + "step": 114, + "teaching_rms": 385539.8513420409, + "traffic_rms": 412564.8150930314 + }, + { + "batch_loss": 14.394222259521484, + "innovation_rms": 648870.9859503832, + "instruction_rms": 0.013848540469714826, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 780216192.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.165094849747511e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3053273344.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1088783740997314, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.908302307128906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 762688448.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 171568048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 637531648.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1979530112.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7717429760.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3053273344.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.207454800605774, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.908204078674316, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 762688448.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 711515.567551978, + "step": 115, + "teaching_rms": 648870.9859503832, + "traffic_rms": 711515.5675519779 + }, + { + "batch_loss": 10.29831600189209, + "innovation_rms": 459115.78023211646, + "instruction_rms": 0.017301107818951562, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2208774144.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3885260876597114e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4685729280.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1791319847106934, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.702535629272461, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 935917504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 248977856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 774098048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2783797248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8042670080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4685729280.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.277698516845703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.702437400817871, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 935917504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 990452.3204729176, + "step": 116, + "teaching_rms": 459115.78023211646, + "traffic_rms": 990452.3204729174 + }, + { + "batch_loss": 15.983086585998535, + "innovation_rms": 743648.727598644, + "instruction_rms": 0.016999510631115787, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2701984256.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5095095062685024e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8142230528.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0982437133789062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.577056884765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1737712896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 351116672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1021388032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3715050496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9312532480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8142230528.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.196800470352173, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.577115058898926, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1737712896.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1318353.248092619, + "step": 117, + "teaching_rms": 743648.727598644, + "traffic_rms": 1318353.248092619 + }, + { + "batch_loss": 21.455076217651367, + "innovation_rms": 459654.07160594675, + "instruction_rms": 0.017904285123425498, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1825057536.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.535535526588753e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6732882944.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8971099853515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.42091178894043, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2411001088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 466172960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1150562944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4610750464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8957001728.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6732882944.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.995656967163086, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.420815467834473, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2411001088.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1261238.6337923594, + "step": 118, + "teaching_rms": 459654.07160594675, + "traffic_rms": 1261238.6337923591 + }, + { + "batch_loss": 30.978700637817383, + "innovation_rms": 515929.26795933314, + "instruction_rms": 0.02559926391591796, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3624651776.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.249718678117071e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6537115648.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5880537033081055, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.863726615905762, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2970498048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 594481792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1283088128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5475331072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8645804032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6537115648.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.686591148376465, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.863544464111328, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2970498048.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1570198.882303244, + "step": 119, + "teaching_rms": 515929.26795933314, + "traffic_rms": 1570198.8823032435 + }, + { + "batch_loss": 46.313987731933594, + "innovation_rms": 887674.3816609307, + "instruction_rms": 0.05430553252162968, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6016745472.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.177944519134578e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11575513088.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.187067031860352, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.435348510742188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3278673408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 751778176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1572963584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6400993280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9256620032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11575513088.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.285594463348389, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.435449600219727, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3278673408.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2018116.5998606002, + "step": 120, + "teaching_rms": 887674.3816609307, + "traffic_rms": 2018116.5998605995 + }, + { + "batch_loss": 68.48441314697266, + "innovation_rms": 2423242.133828464, + "instruction_rms": 0.08466137236765975, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8911654912.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.640921406637328e+21, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20477030400.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.704336643218994, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.067553520202637, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3926031360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1000786048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2490078464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8192696320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17917032448.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20477030400.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.802854061126709, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.067654609680176, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3926031360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2905083.482674749, + "step": 121, + "teaching_rms": 2423242.133828464, + "traffic_rms": 2905083.482674748 + }, + { + "batch_loss": 94.10712432861328, + "innovation_rms": 927861.5632435153, + "instruction_rms": 0.07940597211555883, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12453816320.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.027062447438617e+22, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17780881408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.138214111328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.63764762878418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5331730432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1266778880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2659928320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9911215104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17185187840.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17780881408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.236721515655518, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.63739776611328, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5331730432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3371787.645346142, + "step": 122, + "teaching_rms": 927861.5632435153, + "traffic_rms": 3371787.645346141 + }, + { + "batch_loss": 123.85702514648438, + "innovation_rms": 746895.4544812613, + "instruction_rms": 0.06211133965087674, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17198860288.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.458141417956467e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15838667776.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.3235931396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.714391708374023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6915597312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1539940224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2731613952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11512990720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16017755136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15838667776.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.422091007232666, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.714141845703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6915597312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4035816.0955909397, + "step": 123, + "teaching_rms": 746895.4544812613, + "traffic_rms": 4035816.095590939 + }, + { + "batch_loss": 132.04209899902344, + "innovation_rms": 757328.4332480276, + "instruction_rms": 0.17060467628615014, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32478928896.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.029804126639399e+23, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14129729536.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.455913066864014, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.98267364501953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8328570368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1820896640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2809564672.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13002385408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14893945856.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14129729536.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.55440092086792, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.982437133789062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8328570368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4683946.635323234, + "step": 124, + "teaching_rms": 757328.4332480276, + "traffic_rms": 4683946.63532323 + }, + { + "batch_loss": 173.99708557128906, + "innovation_rms": 958206.5787179807, + "instruction_rms": 0.047042386032286596, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 56410542080.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1307560842776221e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12583766016.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.349010467529297, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.03763771057129, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9586947072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2116201728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2953050624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14411431936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14090462208.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12583766016.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.447488784790039, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.03752899169922, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 9586947072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4942861.103139535, + "step": 125, + "teaching_rms": 958206.5787179807, + "traffic_rms": 4942861.103139535 + }, + { + "batch_loss": 132.0485382080078, + "innovation_rms": 829311.3732660563, + "instruction_rms": 0.07570698215714045, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 77501718528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2691913463518661e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11231479808.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.661961555480957, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.915165901184082, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10710094848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2419495680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3032939264.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15724210176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13127777280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11231479808.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.760429859161377, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.915079116821289, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10710094848.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5867215.073563152, + "step": 126, + "teaching_rms": 829311.3732660563, + "traffic_rms": 5867215.073563152 + }, + { + "batch_loss": 62.65093231201172, + "innovation_rms": 955262.7040685713, + "instruction_rms": 0.03957943711942558, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104671854592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.540158177972399e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10698897408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.914935111999512, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.981043815612793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11710522368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2735220992.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3157251840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16960014336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12358039552.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10698897408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.013393402099609, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.980860710144043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11710522368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6307426.59131226, + "step": 127, + "teaching_rms": 955262.7040685713, + "traffic_rms": 6307426.59131226 + }, + { + "batch_loss": 99.70709991455078, + "innovation_rms": 899231.4958711464, + "instruction_rms": 0.04494638894263172, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 137479749632.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.016385101435236e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10141442048.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.038448333740234, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.423410415649414, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12604200960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3059614720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3243936256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18115268608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11552542720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10141442048.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.136897087097168, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.4234676361084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12604200960.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6757773.792584954, + "step": 128, + "teaching_rms": 899231.4958711464, + "traffic_rms": 6757773.792584953 + }, + { + "batch_loss": 124.09397888183594, + "innovation_rms": 991193.2088844075, + "instruction_rms": 0.0438422637212652, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 161227538432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.152712606826276e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9569802240.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.954068183898926, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.600086212158203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13404343296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3398348288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3387336448.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19201216512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10859485184.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9569802240.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.052507400512695, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.599763870239258, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 13404343296.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6996539.020772765, + "step": 129, + "teaching_rms": 991193.2088844075, + "traffic_rms": 6996539.020772764 + }, + { + "batch_loss": 175.4619598388672, + "innovation_rms": 955734.8888052319, + "instruction_rms": 0.07340491410226559, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 194681864192.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.184639913289457e+24, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9046400000.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.732879161834717, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.996719360351562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14120265728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3746827776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3484793856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20218066944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10168513536.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9046400000.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.831308364868164, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.996397018432617, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14120265728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7774643.3963497775, + "step": 130, + "teaching_rms": 955734.8888052319, + "traffic_rms": 7774643.3963497775 + }, + { + "batch_loss": 192.10887145996094, + "innovation_rms": 1144934.3905645811, + "instruction_rms": 0.058311207998288185, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 244678672384.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0174617594346765e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730277888.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.276386737823486, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.833864212036133, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14835343360.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4111637248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3648094976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21188536320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9704686592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730277888.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.374805927276611, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.834114074707031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14835343360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8118873.733464462, + "step": 131, + "teaching_rms": 1144934.3905645811, + "traffic_rms": 8118873.733464462 + }, + { + "batch_loss": 224.78089904785156, + "innovation_rms": 1260243.1248258296, + "instruction_rms": 0.12756727653457922, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 301314605056.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1157683785365904e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560453120.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.608907699584961, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.39414978027344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15691388928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4485547520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3739104512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22126618624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9380825088.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560453120.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.020030975341797, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.39413833618164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 15691388928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8641859.786285032, + "step": 132, + "teaching_rms": 1260243.1248258296, + "traffic_rms": 8641859.78628503 + }, + { + "batch_loss": 142.56260681152344, + "innovation_rms": 1544357.8587628193, + "instruction_rms": 0.033668443458392996, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 342851256320.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4115716042991618e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8591589376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.939157009124756, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.80146026611328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16550547456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4889611264.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4040637440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23057360896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9307424768.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8591589376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.362827301025391, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.80144500732422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16550547456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9374725.279184673, + "step": 133, + "teaching_rms": 1544357.8587628193, + "traffic_rms": 9374725.279184673 + }, + { + "batch_loss": 115.91201782226562, + "innovation_rms": 1717813.4141861023, + "instruction_rms": 0.02527555869933665, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 362464018432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7845186526093846e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822384640.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.130197525024414, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.4002742767334, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17432786944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5328439296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4388280320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24002050048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9446887424.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822384640.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.553826332092285, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.400259017944336, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17432786944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10144862.431924447, + "step": 134, + "teaching_rms": 1717813.4141861023, + "traffic_rms": 10144862.431924447 + }, + { + "batch_loss": 99.8729476928711, + "innovation_rms": 1980390.060072827, + "instruction_rms": 0.02274017776556502, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 385018757120.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1706472830660785e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9148883968.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.159863471984863, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.443614959716797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18347675648.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5804787712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4763481600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24972541952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9704928256.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9148883968.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.583450317382812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.443599700927734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 18347675648.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10885528.203581817, + "step": 135, + "teaching_rms": 1980390.060072827, + "traffic_rms": 10885528.203581817 + }, + { + "batch_loss": 75.49402618408203, + "innovation_rms": 2384457.772042035, + "instruction_rms": 0.009359169559752148, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 407008083968.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5555914019689393e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9513179136.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.336997032165527, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.792129516601562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19298994176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6319975424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5151878656.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25976582144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12220345344.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9513179136.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.350360870361328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.7921142578125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19298994176.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11857882.256058315, + "step": 136, + "teaching_rms": 2384457.772042035, + "traffic_rms": 11857882.256058315 + }, + { + "batch_loss": 73.42327880859375, + "innovation_rms": 2944438.6819317075, + "instruction_rms": 0.009364397634108757, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 431367946240.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.931277561189807e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9920843776.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.47778606414795, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.40789222717285, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20291078144.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6875370496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5553950208.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 27022807040.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19833159680.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9920843776.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.491148948669434, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.40787696838379, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20291078144.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13516190.615362756, + "step": 137, + "teaching_rms": 2944438.6819317075, + "traffic_rms": 13516190.615362756 + }, + { + "batch_loss": 68.58589935302734, + "innovation_rms": 3809453.5596273635, + "instruction_rms": 0.009820095432631758, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 458206904320.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2948529717576827e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10376694784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.403681755065918, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.258956909179688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21328748544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7474197504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6888434688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 28121315328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30696660992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10376694784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.41704273223877, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.258941650390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 21328748544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16100117.356749043, + "step": 138, + "teaching_rms": 3809453.5596273635, + "traffic_rms": 16100117.356749043 + }, + { + "batch_loss": 68.36637115478516, + "innovation_rms": 4340359.872088085, + "instruction_rms": 0.015158865637862281, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 488710733824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.639200649224628e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13036251136.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.134014129638672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.30332374572754, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22432899072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8123812352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9265564672.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 29301274624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38745448448.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13036251136.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.14737319946289, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.30331039428711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22432899072.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19833481.964961916, + "step": 139, + "teaching_rms": 4340359.872088085, + "traffic_rms": 19833481.964961916 + }, + { + "batch_loss": 75.49397277832031, + "innovation_rms": 7214613.092796876, + "instruction_rms": 0.014430011734133203, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 520160215040.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.965799100128151e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19287148544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.688764572143555, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.547508239746094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23605700608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826602496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14090293248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 30571859968.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62667632640.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19287148544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.70212173461914, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.547493934631348, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23605700608.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26629269.30950584, + "step": 140, + "teaching_rms": 7214613.092796876, + "traffic_rms": 26629269.30950584 + }, + { + "batch_loss": 81.98812103271484, + "innovation_rms": 8553492.508812523, + "instruction_rms": 0.015353915892327194, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 552350318592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.275158838110492e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24793145344.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.082901000976562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.05731201171875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 24832210944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9581654016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19655155712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 31926562816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83761340416.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24793145344.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.096256256103516, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.057369232177734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 24832210944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 36688234.845220916, + "step": 141, + "teaching_rms": 8553492.508812523, + "traffic_rms": 36688234.845220916 + }, + { + "batch_loss": 823.2211303710938, + "innovation_rms": 8932570.825463843, + "instruction_rms": 0.523868707263934, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 649366142976.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.041361575540084e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41085050880.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.765409469604492, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.63032150268555, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25939472384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10316371968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25160849408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 37492523008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97189953536.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41085050880.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.778764724731445, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.63074493408203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25939472384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 43626860.08569767, + "step": 142, + "teaching_rms": 8932570.825463843, + "traffic_rms": 43626860.08569767 + }, + { + "batch_loss": 159.91033935546875, + "innovation_rms": 14752795.556571364, + "instruction_rms": 0.038970700505557995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 706266202112.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2191955671082735e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60879093760.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.296693801879883, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.31455993652344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27133300736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12639844352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34870542336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 50643181568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 131506569216.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60879093760.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.310049057006836, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.31498336791992, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27133300736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 58665321.57803858, + "step": 143, + "teaching_rms": 14752795.556571364, + "traffic_rms": 58665321.57803858 + }, + { + "batch_loss": 265.36834716796875, + "innovation_rms": 20587289.871986125, + "instruction_rms": 0.1302085242250332, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 745870196736.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.053700602650988e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97555234816.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.5384464263916, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.06315994262695, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28205326336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 17513388032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48735436800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 69033844736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 183906631680.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97555234816.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.551799774169922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.06291198730469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 28205326336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 83839571.03056993, + "step": 144, + "teaching_rms": 20587289.871986125, + "traffic_rms": 83839571.03056993 + }, + { + "batch_loss": 154.99888610839844, + "innovation_rms": 30569438.34439123, + "instruction_rms": 0.1013550376362801, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 819454017536.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9902608662756955e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74388586496.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.644330978393555, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.00263214111328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35579478016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 24497661952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69842731008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 96266608640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272327639040.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74388586496.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.657682418823242, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.00238037109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35579478016.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 119684260.5276031, + "step": 145, + "teaching_rms": 30569438.34439123, + "traffic_rms": 119684260.5276031 + }, + { + "batch_loss": 108.03142547607422, + "innovation_rms": 44117563.245524764, + "instruction_rms": 0.03695246766983137, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 840569323520.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0034396814105555e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108833193984.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.662721633911133, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.189998626708984, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46462795776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34503290880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100056285184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 136686559232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 404199505920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108833193984.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.676071166992188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.189746856689453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46462795776.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 166110366.14039358, + "step": 146, + "teaching_rms": 44117563.245524764, + "traffic_rms": 166110366.14039358 + }, + { + "batch_loss": 139.3473358154297, + "innovation_rms": 58547582.97072246, + "instruction_rms": 0.03518118831728243, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 856650809344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.064012332588192e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 194040201216.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.656402587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.121780395507812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 49917181952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 48411521024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139082301440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 192537395200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 558508343296.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 194040201216.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.66975212097168, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.12152862548828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 49917181952.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 235778746.96374968, + "step": 147, + "teaching_rms": 58547582.97072246, + "traffic_rms": 235778746.96374968 + }, + { + "batch_loss": 115.96622467041016, + "innovation_rms": 45701824.61498273, + "instruction_rms": 0.02423537194871074, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 869119754240.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.16085589430076e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168613167104.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.62250518798828, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.56891441345215, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 60335673344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 64587255808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 161757364224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 250730790912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 581933924352.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168613167104.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.635854721069336, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.569339752197266, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 60335673344.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 288188806.9391092, + "step": 148, + "teaching_rms": 45701824.61498273, + "traffic_rms": 288188806.9391092 + }, + { + "batch_loss": 94.11609649658203, + "innovation_rms": 53939248.92084282, + "instruction_rms": 0.015872655805843846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 880988323840.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.255178247267053e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156410183680.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.587587356567383, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.695887565612793, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 75976695808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 83397107712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 188098478080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 311492804608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 607620235264.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156410183680.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.600934982299805, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.69631290435791, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 75976695808.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 375990937.7626014, + "step": 149, + "teaching_rms": 53939248.92084282, + "traffic_rms": 375990937.7626014 + }, + { + "batch_loss": 157.3045654296875, + "innovation_rms": 69557572.39264233, + "instruction_rms": 0.018229122343958747, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1088140869632.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.309103614217728e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 215657316352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.554418563842773, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.054739952087402, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97542430720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 105806577664.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 224094732288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 377436995584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 659442040832.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 215657316352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.567766189575195, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.055164337158203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 97542430720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 470357969.0170351, + "step": 150, + "teaching_rms": 69557572.39264233, + "traffic_rms": 470357969.0170351 + }, + { + "batch_loss": 144.1659698486328, + "innovation_rms": 94206209.21872225, + "instruction_rms": 0.014486149293648037, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1482984914944.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.288957463846228e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 296647491584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.523475646972656, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.644341468811035, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 127207178240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 133531607040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 277250277376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 454070763520.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 766337679360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 296647491584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.536821365356445, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.644765853881836, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 127207178240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 575784362.9037259, + "step": 151, + "teaching_rms": 94206209.21872225, + "traffic_rms": 575784362.9037259 + }, + { + "batch_loss": 189.64889526367188, + "innovation_rms": 135533366.90950778, + "instruction_rms": 0.021934186431885206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1988855726080.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.172422464003577e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 412098658304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.494813919067383, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.567789077758789, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 168417050624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 169508241408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 359766327296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 553210478592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 991397347328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 412098658304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.50815773010254, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.567693710327148, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 168417050624.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 715733962.3653249, + "step": 152, + "teaching_rms": 135533366.90950778, + "traffic_rms": 715733962.3653249 + }, + { + "batch_loss": 132.97732543945312, + "innovation_rms": 188008359.96886423, + "instruction_rms": 0.024260312576210127, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2517190836224.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.941339639823616e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 552127561728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.467985153198242, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.74736785888672, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 223629803520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 217391398912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 478831607808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 690377326592.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1371668611072.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 552127561728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.481327056884766, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.74727249145508, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 223629803520.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 891184974.1423926, + "step": 153, + "teaching_rms": 188008359.96886423, + "traffic_rms": 891184974.1423926 + }, + { + "batch_loss": 223.7243194580078, + "innovation_rms": 295504674.4992325, + "instruction_rms": 0.04310235281130661, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2765494419456.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6660021542736217e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 642949054464.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.442720413208008, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.281253814697266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 287924715520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 284813459456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 674220474368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 910056751104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2196794376192.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 642949054464.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.4560604095459, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.281158447265625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 287924715520.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1196030349.3633492, + "step": 154, + "teaching_rms": 295504674.4992325, + "traffic_rms": 1196030349.3633492 + }, + { + "batch_loss": 189.26058959960938, + "innovation_rms": 273091471.6584243, + "instruction_rms": 0.042800123958121244, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3651054338048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.887892504713052e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1243626340352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.40894889831543, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.911312103271484, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 379682095104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 367961702400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 831482429440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1167940911104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2578842124288.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1243626340352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.422286987304688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.911409378051758, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 379682095104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1595879112.012313, + "step": 155, + "teaching_rms": 273091471.6584243, + "traffic_rms": 1595879112.012313 + }, + { + "batch_loss": 472.86553955078125, + "innovation_rms": 276927998.68854827, + "instruction_rms": 0.18054486602771427, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5090762555392.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.310482810416042e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1098041524224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.07529067993164, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.65423583984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 489486254080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 466392383488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 984306679808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1448163934208.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2802230493184.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1098041524224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.088626861572266, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.654293060302734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 489486254080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1730452468.6401684, + "step": 156, + "teaching_rms": 276927998.68854827, + "traffic_rms": 1730452468.6401684 + }, + { + "batch_loss": 482.73089599609375, + "innovation_rms": 345537422.1483314, + "instruction_rms": 0.14550218686374777, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6841530056704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.61186954743013e+25, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1311777226752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.685768127441406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.09092712402344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 620663996416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 584541274112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1181488775168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1760868564992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3127046307840.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1311777226752.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.6991024017334, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.09098815917969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 620663996416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2125404824.7786186, + "step": 157, + "teaching_rms": 345537422.1483314, + "traffic_rms": 2125404824.7786186 + }, + { + "batch_loss": 400.4479064941406, + "innovation_rms": 562670223.3602356, + "instruction_rms": 0.10165352824202659, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9409615888384.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1836045274269703e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862440845312.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.265859603881836, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.964447021484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 806908067840.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 738975350784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1544340897792.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2178708275200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4178397102080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862440845312.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.279193878173828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.964508056640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 806908067840.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2774666702.075097, + "step": 158, + "teaching_rms": 562670223.3602356, + "traffic_rms": 2774666702.075097 + }, + { + "batch_loss": 455.4384765625, + "innovation_rms": 714863927.6250032, + "instruction_rms": 0.06889830864125245, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11713114013696.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3794558482444763e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1883317338112.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.815580368041992, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.722476959228516, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 995239788544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 937413705728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1984383680512.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2730596499456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5518883028992.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1883317338112.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.82891273498535, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.722726821899414, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 995239788544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3754580071.8155904, + "step": 159, + "teaching_rms": 714863927.6250032, + "traffic_rms": 3754580071.8155904 + }, + { + "batch_loss": 533.408203125, + "innovation_rms": 1123342880.9937737, + "instruction_rms": 0.07828943669832045, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14483343802368.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8451013157841066e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2060317622272.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.840742111206055, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.25597381591797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1185136508928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1209794035712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2723802775552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3580942352384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8503458529280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2060317622272.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.854074478149414, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.25606918334961, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1185136508928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4847868420.850484, + "step": 160, + "teaching_rms": 1123342880.9937737, + "traffic_rms": 4847868420.850484 + }, + { + "batch_loss": 441.4041442871094, + "innovation_rms": 1054908353.8453695, + "instruction_rms": 0.053516112500550383, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 18482836013056.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.3010865923872166e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432967045120.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.663524627685547, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.04441833496094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1528433213440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1544255307776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3344612458496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4576534593536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9955922935808.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432967045120.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.676856994628906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.04451370239258, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1528433213440.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5444431131.368202, + "step": 161, + "teaching_rms": 1054908353.8453695, + "traffic_rms": 5444431131.368202 + }, + { + "batch_loss": 384.3792724609375, + "innovation_rms": 1807944938.826639, + "instruction_rms": 0.049445302624342956, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 23441222664192.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.462350872224125e+26, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6356056670208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.454090118408203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.12751770019531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2164038828032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1997534134272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4532788527104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6017054670848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14405200773120.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6356056670208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.46742057800293, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.12760925292969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2164038828032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7670753447.155128, + "step": 162, + "teaching_rms": 1807944938.826639, + "traffic_rms": 7670753447.155128 + }, + { + "batch_loss": 470.5538024902344, + "innovation_rms": 844471973.3220799, + "instruction_rms": 0.05150181935052064, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31218353569792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.580916429622273e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5924225286144.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.0305233001709, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.808868408203125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2756461461504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2473983737856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4764496560128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7388228222976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13711737618432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5924225286144.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.043851852416992, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.808616638183594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2756461461504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10636934976.655996, + "step": 163, + "teaching_rms": 844471973.3220799, + "traffic_rms": 10636934976.655996 + }, + { + "batch_loss": 554.2662963867188, + "innovation_rms": 721274347.2014536, + "instruction_rms": 0.06794942417321895, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 40755993247744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.505699058244424e+27, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5514479534080.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.546422958374023, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.319114685058594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3307909414912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2960554459136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4865707212800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8660468105216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12722399870976.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5514479534080.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.559749603271484, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.31888198852539, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3307909414912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13072862970.800821, + "step": 164, + "teaching_rms": 721274347.2014536, + "traffic_rms": 13072862970.800821 + }, + { + "batch_loss": 729.2855834960938, + "innovation_rms": 756227756.6492562, + "instruction_rms": 0.07614749954470469, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 51655433256960.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3108070985893555e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5142383427584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.948389053344727, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.449493408203125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3822147862528.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3458627796992.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4980732854272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9838644953088.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11781771624448.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5142383427584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.961715698242188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.44968032836914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3822147862528.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15470660304.449001, + "step": 165, + "teaching_rms": 756227756.6492562, + "traffic_rms": 15470660304.449001 + }, + { + "batch_loss": 566.7844848632812, + "innovation_rms": 889869325.9197832, + "instruction_rms": 0.10426577314025894, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 63684890066944.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4053241454235018e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4834862825472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.816198348999023, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.74602127075195, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4305634197504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3977817620480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5191898234880.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10938847592448.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11002025345024.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4834862825472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.403112411499023, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.74570083618164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4305634197504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17751096348.899776, + "step": 166, + "teaching_rms": 889869325.9197832, + "traffic_rms": 17751096348.899776 + }, + { + "batch_loss": 921.3631591796875, + "innovation_rms": 1004316493.2630985, + "instruction_rms": 0.07196254912588727, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 76472052088832.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.79976954902761e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4589555286016.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.86836814880371, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.40864562988281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4764589883392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4526256422912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5484389597184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11975873527808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10370255159296.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4589555286016.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.44614028930664, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.40870666503906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4764589883392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19937712021.774593, + "step": 167, + "teaching_rms": 1004316493.2630985, + "traffic_rms": 19937712021.774593 + }, + { + "batch_loss": 1240.41015625, + "innovation_rms": 1148929329.7937045, + "instruction_rms": 0.13434476308514184, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 90399070748672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.373039165421784e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4421627936768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.91276741027832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.22900390625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5206752886784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5114298892288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5880424693760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12965324521472.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9894508888064.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4421627936768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.490583419799805, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.22889709472656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5206752886784.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22023317111.182037, + "step": 168, + "teaching_rms": 1148929329.7937045, + "traffic_rms": 22023317111.182037 + }, + { + "batch_loss": 1426.1195068359375, + "innovation_rms": 1436183334.460174, + "instruction_rms": 0.1816562763061736, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104662279651328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.092733310403704e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4337254006784.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.950347900390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86.47278594970703, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5640478523392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5762138505216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6478396129280.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13941241544704.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9759173378048.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4337254006784.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.52820587158203, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86.47269439697266, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5640478523392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24144093444.473637, + "step": 169, + "teaching_rms": 1436183334.460174, + "traffic_rms": 24144093444.473637 + }, + { + "batch_loss": 4356.41748046875, + "innovation_rms": 1672995656.035789, + "instruction_rms": 0.6869893410468821, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 119685093785600.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.819881710878834e+28, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4348250947584.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.734716415405273, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206.3881378173828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6075303591936.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6484127645696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7219889307648.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14928901570560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9876596064256.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4348250947584.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.677310943603516, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206.38824462890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6075303591936.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26326302708.68871, + "step": 170, + "teaching_rms": 1672995656.035789, + "traffic_rms": 26326302708.68871 + }, + { + "batch_loss": 4019.689453125, + "innovation_rms": 1795661070.51393, + "instruction_rms": 0.11544781460947937, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 133378137391104.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0035341396759162e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4163561848832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.160646438598633, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158.3726348876953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6491659567104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7257093832704.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7729660821504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15934452400128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10055512489984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4163561848832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.062246322631836, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 158.3727569580078, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6491659567104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28660706288.17737, + "step": 171, + "teaching_rms": 1795661070.51393, + "traffic_rms": 28660706288.17737 + }, + { + "batch_loss": 6887.56640625, + "innovation_rms": 2449246965.371679, + "instruction_rms": 0.08525473372796885, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 149425703878656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1615298828566376e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4545301184512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.072120666503906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273.9119873046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6881840463872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8119573282816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8624793976832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17056583909376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11221312995328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4545301184512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.973731994628906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273.9117431640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6881840463872.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30888400009.469913, + "step": 172, + "teaching_rms": 2449246965.371679, + "traffic_rms": 30888400009.469913 + }, + { + "batch_loss": 12747.369140625, + "innovation_rms": 2540060226.6986513, + "instruction_rms": 0.18046192744163317, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 167239449837568.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5732295707263996e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4993330970624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.12508773803711, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 306.4119567871094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7234149941248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9055199297536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9356259098624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18274232303616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12176478699520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4993330970624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.02670669555664, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 306.4117126464844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7234149941248.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33401443992.39286, + "step": 173, + "teaching_rms": 2540060226.6986513, + "traffic_rms": 33401443992.39286 + }, + { + "batch_loss": 12141.966796875, + "innovation_rms": 3344446352.078253, + "instruction_rms": 0.9657868557972595, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 224905425584128.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.317482854839132e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7393406615552.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.78248596191406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 297.3453369140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7551155961856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 10113709506560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10585097895936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19743595560960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14693623136256.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7393406615552.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.0313720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 297.3450012207031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7551155961856.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35943314024.66915, + "step": 174, + "teaching_rms": 3344446352.078253, + "traffic_rms": 35943314024.66915 + }, + { + "batch_loss": 16439.427734375, + "innovation_rms": 4119369815.6098495, + "instruction_rms": 0.1735758189917861, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 330412068962304.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.0737366317055985e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9915418542080.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.68596267700195, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280.0301208496094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7836375449600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11337440690176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12237314981888.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21608259387392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18646644555776.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9915418542080.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.9348258972168, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280.03033447265625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7836375449600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 39652610071.181366, + "step": 175, + "teaching_rms": 4119369815.6098495, + "traffic_rms": 39652610071.181366 + }, + { + "batch_loss": 19506.34765625, + "innovation_rms": 4379462225.297854, + "instruction_rms": 0.08050733368297947, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 450463988187136.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.866275576516627e+29, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12241595269120.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.58269119262695, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 305.19842529296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8092875489280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 12732148482048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13947076870144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23839494897664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22312353005568.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12241595269120.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.81597137451172, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 305.19854736328125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8092875489280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 43217401420.27285, + "step": 176, + "teaching_rms": 4379462225.297854, + "traffic_rms": 43217401420.27285 + }, + { + "batch_loss": 22580.67578125, + "innovation_rms": 8188135059.238234, + "instruction_rms": 0.19934854279455136, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 597755294646272.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3211250160072433e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13848484511744.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.75733947753906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 225.56295776367188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8319342739456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 14582483517440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18503350353920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 27790277607424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39507827097600.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13848484511744.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.99059295654297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 225.5630645751953, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8319342739456.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 46482487155.66698, + "step": 177, + "teaching_rms": 8188135059.238234, + "traffic_rms": 46482487155.66698 + }, + { + "batch_loss": 16529.125, + "innovation_rms": 9217218789.504635, + "instruction_rms": 0.71846375794798, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 810591123931136.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.401263046316495e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27408306536448.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.12248229980469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 185.61013793945312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8530118049792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16949584068608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23671007608832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 33469449109504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56791723409408.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27408306536448.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.02622985839844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 185.6102294921875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8530118049792.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 46134051531.71246, + "step": 178, + "teaching_rms": 9217218789.504635, + "traffic_rms": 46134051531.71246 + }, + { + "batch_loss": 23942.150390625, + "innovation_rms": 14483242150.607914, + "instruction_rms": 0.8734136702711126, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1022020015882240.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.263970040617879e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26661712035840.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.4560317993164, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.5872802734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8873131900928.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20097248788480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31476653490176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 43117231210496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 96477833592832.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26661712035840.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.68923950195312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.587158203125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8873131900928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 65908574074.52317, + "step": 179, + "teaching_rms": 14483242150.607914, + "traffic_rms": 65908574074.52317 + }, + { + "batch_loss": 21855.734375, + "innovation_rms": 11500349981.732698, + "instruction_rms": 0.2717812699026589, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1315253069545472.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.615413888465897e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34812119220224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.21955108642578, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.8658752441406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10486685892608.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 23744858816512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36476096086016.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 53850799079424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107335687077888.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34812119220224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.4527359008789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.86578369140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 10486685892608.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 85046629033.2337, + "step": 180, + "teaching_rms": 11500349981.732698, + "traffic_rms": 85046629033.2337 + }, + { + "batch_loss": 13103.4482421875, + "innovation_rms": 11367487950.422092, + "instruction_rms": 0.3134198896436296, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1804282340835328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.978624362431952e+30, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37827123347456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.97914123535156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239.52664184570312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12894084268032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 27712278233088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39674198360064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 65067135533056.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112163356147712.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37827123347456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.07742309570312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239.52658081054688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 12894084268032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 98881259590.69688, + "step": 181, + "teaching_rms": 11367487950.422092, + "traffic_rms": 98881259590.69688 + }, + { + "batch_loss": 21886.498046875, + "innovation_rms": 5755759673.007816, + "instruction_rms": 5.138533273818843, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2530127384674304.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3621007104150074e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35326219255808.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.63927459716797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.0321044921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16297713729536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31575685201920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38634061299712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 75416085725184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103489501921280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35326219255808.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.5322036743164, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.0320281982422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 16297713729536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 109046581376.66664, + "step": 182, + "teaching_rms": 5755759673.007816, + "traffic_rms": 109046581376.66664 + }, + { + "batch_loss": 3457.849609375, + "innovation_rms": 18690496743.452408, + "instruction_rms": 0.09941999911643457, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3119193423609856.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7322751261994084e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47288252628992.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.5296859741211, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.90118408203125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17760346177536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 36275545440256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46998610771968.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 87878268878848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124621797982208.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47288252628992.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.4226303100586, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.90110778808594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17760346177536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 128889804069.91342, + "step": 183, + "teaching_rms": 18690496743.452408, + "traffic_rms": 128889804069.91342 + }, + { + "batch_loss": 4664.59130859375, + "innovation_rms": 25242082874.235085, + "instruction_rms": 0.13321625038287602, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3713426308202496.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.959482805628578e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41314330607616.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.71575927734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.34815979003906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19993154551808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 42295416061952.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60198702022656.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 104128663519232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162503912849408.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41314330607616.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.60871124267578, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.34808349609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 19993154551808.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 158474069059.66406, + "step": 184, + "teaching_rms": 25242082874.235085, + "traffic_rms": 158474069059.66406 + }, + { + "batch_loss": 3239.353271484375, + "innovation_rms": 31194436574.511723, + "instruction_rms": 0.06739302864521944, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4463843799465984.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.285525973655177e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74986035347456.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 88.11970520019531, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.97099304199219, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22284009996288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 49977661325312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76822435856384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 125280488259584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 211518280957952.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74986035347456.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 88.0126724243164, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.97090911865234, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22284009996288.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 181864254017.7989, + "step": 185, + "teaching_rms": 31194436574.511723, + "traffic_rms": 181864254017.7989 + }, + { + "batch_loss": 1741.1534423828125, + "innovation_rms": 39455560494.08645, + "instruction_rms": 0.14700276361099046, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5453486942584832.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.148749753393088e+31, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56283294072832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.34422302246094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.39231872558594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22493479829504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 59940383227904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99627210637312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 153194554458112.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 279140661985280.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56283294072832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.2372055053711, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.3922348022461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22493479829504.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 207500029042.8131, + "step": 186, + "teaching_rms": 39455560494.08645, + "traffic_rms": 207500029042.8131 + }, + { + "batch_loss": 2199.78515625, + "innovation_rms": 46673113210.08662, + "instruction_rms": 0.056276949476680646, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6708413572579328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3744352113029207e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87300579000320.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.47895050048828, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 92.33809661865234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30223999434752.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 72500012646400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125596327739392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 188487173144576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 352926220419072.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87300579000320.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.37194061279297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 92.3380126953125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30223999434752.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 248096709038.15817, + "step": 187, + "teaching_rms": 46673113210.08662, + "traffic_rms": 248096709038.15817 + }, + { + "batch_loss": 2173.311279296875, + "innovation_rms": 59038578516.92875, + "instruction_rms": 0.041984723720431376, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9669285643288576.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1741366760233188e+32, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78806266150912.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.48372650146484, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.08622741699219, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32665575096320.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 88199795834880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156997865439232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 233438250532864.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 449510807437312.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 78806266150912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.37672424316406, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.08614349365234, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 32665575096320.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 319977577002.8028, + "step": 188, + "teaching_rms": 59038578516.92875, + "traffic_rms": 319977577002.8028 + }, + { + "batch_loss": 1636.298828125, + "innovation_rms": 74247736458.641, + "instruction_rms": 0.0364093607983961, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4351491485663232e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.82519039882843e+31, + "nonfinite_indices": [ + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61322259595264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.69876861572266, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.34724426269531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38021831327744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 107919223816192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197194313367552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 290836495466496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 573982382227456.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61322259595264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.5917739868164, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.34716033935547, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38021831327744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 381656627232.8912, + "step": 189, + "teaching_rms": 74247736458.641, + "traffic_rms": 381656627232.8912 + }, + { + "batch_loss": 1950.2254638671875, + "innovation_rms": 98932932396.85075, + "instruction_rms": 0.040281365375289695, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.036613147472691e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2148820520567721e+32, + "nonfinite_indices": [ + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99184921280512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.23914337158203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.15935516357422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46207577620480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 133106824642560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 251875974709248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 367057401020416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 762208921321472.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99184921280512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.3353042602539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.15926742553711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46207577620480.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 479529178227.8159, + "step": 190, + "teaching_rms": 98932932396.85075, + "traffic_rms": 479529178227.8159 + }, + { + "batch_loss": 1566.569091796875, + "innovation_rms": 113471849959.32155, + "instruction_rms": 0.024001101981770284, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8805009919442944e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.6143944136679006e+32, + "nonfinite_indices": [ + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 447177935552512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 86.79873657226562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.59592819213867, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 81422731706368.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 163664225107968.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 305573937545216.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 460866398978048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 938089845358592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 447177935552512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 86.89488983154297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.59584426879883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 81422731706368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 633000614532.5641, + "step": 191, + "teaching_rms": 113471849959.32155, + "traffic_rms": 633000614532.5641 + }, + { + "batch_loss": 1453.3660888671875, + "innovation_rms": 20958139843.96959, + "instruction_rms": 0.02412318972955573, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6001010728042496e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.4039133441676908e+31, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 402451991625728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.7978515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.86890411376953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 121667934224384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 192582290243584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 289180651356160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 545294483718144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 844280847400960.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 402451991625728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.89398956298828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.86902618408203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 121667934224384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 296454360829.051, + "step": 192, + "teaching_rms": 20958139843.96959, + "traffic_rms": 296454360829.051 + }, + { + "batch_loss": 1712.7872314453125, + "innovation_rms": 26201493087.62711, + "instruction_rms": 0.05840958012465519, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.490197611420058e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.770419622054538e+31, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 362194592071680.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.02295684814453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.25102996826172, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 157887393431552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 220275232735232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 276929357807616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 621279770050560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 759852729106432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 362194592071680.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.11908721923828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.25094223022461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 157887393431552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 348919577062.3574, + "step": 193, + "teaching_rms": 26201493087.62711, + "traffic_rms": 348919577062.3574 + }, + { + "batch_loss": 1872.2052001953125, + "innovation_rms": 34812260739.78756, + "instruction_rms": 0.03583603329245215, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.537342420942848e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.236718268500719e+31, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 325959328727040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.73896789550781, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.998558044433594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 190483326304256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 247105222344704.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 268299845763072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 689666521038848.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 683867442774016.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 325959328727040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.83509063720703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.99845886230469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 190483326304256.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 399107926946.89154, + "step": 194, + "teaching_rms": 34812260739.78756, + "traffic_rms": 399107926946.89154 + }, + { + "batch_loss": 1437.83935546875, + "innovation_rms": 30679755676.37481, + "instruction_rms": 0.027852677908370566, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.728940724525466e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.327216959119049e+31, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 293344320159744.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 103.07122039794922, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.6498031616211, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 219817751609344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 273390740963328.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 262855253295104.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 751214610350080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 615480691785728.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 293344320159744.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 103.1673355102539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.64970397949219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 219817751609344.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 448948835850.5411, + "step": 195, + "teaching_rms": 30679755676.37481, + "traffic_rms": 448948835850.5411 + }, + { + "batch_loss": 1374.138916015625, + "innovation_rms": 32387438440.233013, + "instruction_rms": 0.025310505746109796, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.693354827507302e+16, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.7734603274517118e+32, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 263987899924480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 106.03675079345703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.07857894897461, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 246216549990400.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 299413746483200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 260229954535424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 806607877308416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 553932602474496.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 263987899924480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 106.13285064697266, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.07847595214844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 246216549990400.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 497424212906.8435, + "step": 196, + "teaching_rms": 32387438440.233013, + "traffic_rms": 497424212906.8435 + }, + { + "batch_loss": 1273.957275390625, + "innovation_rms": 38439535182.68601, + "instruction_rms": 0.03997890548678539, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1547666397370778e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.8932842853513424e+32, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 237564472590336.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.48097229003906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.05542755126953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 269973003960320.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 325425679040512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 260119308795904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 856461844414464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 498539335516160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 237564472590336.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.57706451416016, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.055747985839844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 269973003960320.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 545550243028.5411, + "step": 197, + "teaching_rms": 38439535182.68601, + "traffic_rms": 545550243028.5411 + }, + { + "batch_loss": 1093.5308837890625, + "innovation_rms": 40263447184.00676, + "instruction_rms": 0.029601850812468366, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4613138316355174e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.605034495594818e+32, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213781024079872.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.67375183105469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.97183609008789, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 291351119790080.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 351652762615808.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 262270902861824.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 901330361122816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 448685401964544.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213781024079872.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.76983642578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.97216033935547, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 291351119790080.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 593635776853.8279, + "step": 198, + "teaching_rms": 40263447184.00676, + "traffic_rms": 593635776853.8279 + }, + { + "batch_loss": 804.0917358398438, + "innovation_rms": 49088469837.99691, + "instruction_rms": 0.04315092565413195, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7867110337006797e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.756966009019718e+32, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206117930008576.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.63233184814453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.258270263671875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 310588513189888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 378300484550656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266477118685184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 941712046292992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 403816851701760.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206117930008576.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.72840881347656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.25859451293945, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 310588513189888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 642471012628.6772, + "step": 199, + "teaching_rms": 49088469837.99691, + "traffic_rms": 642471012628.6772 + }, + { + "batch_loss": 1012.3644409179688, + "innovation_rms": 58971424698.01335, + "instruction_rms": 0.04771545650080209, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1222483099543142e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.3812049831218184e+33, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249888327598080.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.24915313720703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.21766662597656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 327899043332096.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 405557387198464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 272568909037568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 978055589789696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 363435166531584.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249888327598080.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.34522247314453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.2176513671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 327899043332096.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 692472762002.3455, + "step": 200, + "teaching_rms": 58971424698.01335, + "traffic_rms": 692472762002.3455 + }, + { + "batch_loss": 649.5167846679688, + "innovation_rms": 67572646513.58717, + "instruction_rms": 0.03893689807604019, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4679029829690982e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.7389244696638594e+33, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 307750714212352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.69856262207031, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.587745666503906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 343475245547520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 433598423367680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 280410462355456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010764785647616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 425048317886464.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 307750714212352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.79462432861328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.5877685546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 343475245547520.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 743924161926.1117, + "step": 201, + "teaching_rms": 67572646513.58717, + "traffic_rms": 743924161926.1117 + }, + { + "batch_loss": 542.8773193359375, + "innovation_rms": 92569535984.37682, + "instruction_rms": 0.04298582871103677, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8136895973792154e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.0435499458609897e+33, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410251283136512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.96442413330078, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.26601791381836, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 357490394923008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 462587875557376.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 289894589005824.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1040203028365312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 599357854318592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410251283136512.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.06047821044922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.26604080200195, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 357490394923008.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 798527883753.8, + "step": 202, + "teaching_rms": 92569535984.37682, + "traffic_rms": 798527883753.8 + }, + { + "batch_loss": 663.7919311523438, + "innovation_rms": 141646882802.97342, + "instruction_rms": 0.05287138672373288, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.163733039775744e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.2831380756991938e+33, + "nonfinite_indices": [ + 3, + 4 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 555095666196480.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.85179901123047, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.082155227661133, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 370100452458496.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 492681805430784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 300939164516352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1066697473654784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1033710312882176.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 555095666196480.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.94784545898438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.082178115844727, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 370100452458496.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 858926084647.9165, + "step": 203, + "teaching_rms": 141646882802.97342, + "traffic_rms": 858926084647.9165 + }, + { + "batch_loss": 477.0784606933594, + "innovation_rms": 193522239129.12103, + "instruction_rms": 0.04011928698325977, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.506393776987832e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.475844487794631e+33, + "nonfinite_indices": [ + 3, + 4, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 583479091789824.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.6782455444336, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.667213439941406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 381445809897472.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 524030167744512.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 313483723800576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1090542460993536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1690136739840000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 583479091789824.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.77427673339844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.667236328125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 381445809897472.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 925477437994.8357, + "step": 204, + "teaching_rms": 193522239129.12103, + "traffic_rms": 925477437994.8357 + }, + { + "batch_loss": 394.3440856933594, + "innovation_rms": 268867144051.73404, + "instruction_rms": 0.03628403599902692, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.843184840094515e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.974714516586011e+33, + "nonfinite_indices": [ + 3, + 4, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 890195591299072.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.4638442993164, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.780467987060547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 430932288864256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 556778857168896.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 327486827134976.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1112002936176640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2641077664219136.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 890195591299072.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.55986785888672, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.78049087524414, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 430932288864256.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1001678429192.4694, + "step": 205, + "teaching_rms": 268867144051.73404, + "traffic_rms": 1001678429192.4694 + }, + { + "batch_loss": 929.8425903320312, + "innovation_rms": 57426565032.84972, + "instruction_rms": 0.033881783330635816, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1720425832094106e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.559520744448222e+30, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 801132934856704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.102783203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.174591064453125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 511045575639040.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 591071251791872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 342923812012032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1152782107148288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2376969924640768.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 801132934856704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.19879913330078, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.17441177368164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 511045575639040.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1006567133047.6594, + "step": 206, + "teaching_rms": 57426565032.84972, + "traffic_rms": 1006567133047.6594 + }, + { + "batch_loss": 851.6187133789062, + "innovation_rms": 61447799729.39615, + "instruction_rms": 0.07848570653021676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.492310048134922e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1041046974991231e+31, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 720968511127552.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.54888916015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.55281829833984, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 583142406619136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 627049723068416.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 359784880537600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1366709462106112.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2139272878489600.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 720968511127552.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.64488983154297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.55264282226562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 583142406619136.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1069033724797.1064, + "step": 207, + "teaching_rms": 61447799729.39615, + "traffic_rms": 1069033724797.1064 + }, + { + "batch_loss": 1293.8623046875, + "innovation_rms": 66086685059.20164, + "instruction_rms": 0.04431163407968849, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.790935344654582e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.650183381096223e+31, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 648813328990208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.87081909179688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.65199279785156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 648023726096384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 664857112215552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 378073555927040.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1559243987615744.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1925345523531776.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 648813328990208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.96681213378906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.65182113647461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 648023726096384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1134592867937.3794, + "step": 208, + "teaching_rms": 66086685059.20164, + "traffic_rms": 1134592867937.3794 + }, + { + "batch_loss": 1341.189453125, + "innovation_rms": 70184754757.94109, + "instruction_rms": 0.08238059712113063, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.077315861212037e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.122873255672961e+31, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 583867182350336.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.95050048828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.622257232666016, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 706410451042304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 704637636182016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 397805373882368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1732525114261504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1732810863804416.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 583867182350336.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.0464859008789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37.62208557128906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 706410451042304.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1203411365155.412, + "step": 209, + "teaching_rms": 70184754757.94109, + "traffic_rms": 1203411365155.412 + }, + { + "batch_loss": 1450.205078125, + "innovation_rms": 73045799160.39769, + "instruction_rms": 0.05579968820264074, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.365470714658816e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.042490584176644e+32, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 525409825325056.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.86009216308594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.864864349365234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 758951457062912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 746538330488832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 419006741741568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1888478061133824.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1559529737158656.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 525409825325056.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.95606994628906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.86491394042969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 758951457062912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1276297682803.1677, + "step": 210, + "teaching_rms": 73045799160.39769, + "traffic_rms": 1276297682803.1677 + }, + { + "batch_loss": 1719.4906005859375, + "innovation_rms": 77625455972.41933, + "instruction_rms": 0.05567056682351812, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.635638932264714e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.756081296269255e+32, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 472792919179264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.48326110839844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.97013473510742, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 806230725492736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 790709787426816.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 441714233835520.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2028835713318912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1403576790286336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 472792919179264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.5792236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.97018814086914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 806230725492736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1352834413680.395, + "step": 211, + "teaching_rms": 77625455972.41933, + "traffic_rms": 1352834413680.395 + }, + { + "batch_loss": 1493.4110107421875, + "innovation_rms": 82569002808.2167, + "instruction_rms": 0.05139614527553384, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.892880672699187e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1749922583415153e+33, + "nonfinite_indices": [ + 3, + 4, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 425432985894912.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.86290740966797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.90888214111328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 848774054281216.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 837307162689536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 465973752627200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2155157680816128.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1263219138101248.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 425432985894912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.9588623046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.90889358520508, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 848774054281216.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1433431755588.316, + "step": 212, + "teaching_rms": 82569002808.2167, + "traffic_rms": 1433431755588.316 + }, + { + "batch_loss": 1243.0240478515625, + "innovation_rms": 85464551297.11417, + "instruction_rms": 0.025602047958102347, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.131684977525391e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.619938293345805e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 382804797947904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.16415405273438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.21539306640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 887054560919552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 886491182006272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 491839991840768.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2268847344189440.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1136897170604032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 382804797947904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.26010131835938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.21581268310547, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 887054560919552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1515069235383.939, + "step": 213, + "teaching_rms": 85464551297.11417, + "traffic_rms": 1515069235383.939 + }, + { + "batch_loss": 1593.4219970703125, + "innovation_rms": 90500258609.16382, + "instruction_rms": 0.030394142047591072, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.361238266393395e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.087378422726546e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 344435606945792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.65135955810547, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.3890380859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 921498118258688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 938428812230656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 519376168026112.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2371167960694784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1023207440121856.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 344435606945792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.74729919433594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.38886260986328, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 921498118258688.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1604123482101.1584, + "step": 214, + "teaching_rms": 90500258609.16382, + "traffic_rms": 1604123482101.1584 + }, + { + "batch_loss": 3916.453125, + "innovation_rms": 95817390187.14421, + "instruction_rms": 0.04228599695561111, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.571386549420687e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.614921991725575e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309899909136384.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.61689758300781, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.91352081298828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 952488119238656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 993294200864768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 548653584351232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2463256622923776.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 920886689398784.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309899909136384.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.71282196044922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.913352966308594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 952488119238656.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1698121057708.3337, + "step": 215, + "teaching_rms": 95817390187.14421, + "traffic_rms": 1698121057708.3337 + }, + { + "batch_loss": 6261.92578125, + "innovation_rms": 101440550557.80034, + "instruction_rms": 0.04371963177081869, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.774338528844186e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.1970045565803715e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 278814663901184.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.5846939086914, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.28570556640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 980369570529280.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1051269347147776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 579751597047808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2546136338399232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 828798027169792.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 278814663901184.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.68061065673828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.28562927246094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 980369570529280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1797378648198.2385, + "step": 216, + "teaching_rms": 101440550557.80034, + "traffic_rms": 1797378648198.2385 + }, + { + "batch_loss": 1569.0404052734375, + "innovation_rms": 107423244482.05092, + "instruction_rms": 0.027223999858339547, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.95647537876435e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.82862551653068e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 250835166560256.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.55451965332031, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 226.49485778808594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1005453119062016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1112545041580032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 612757078540288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2620728109170688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 745918177476608.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 250835166560256.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.65042877197266, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 226.49478149414062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1005453119062016.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1902228515095.4216, + "step": 217, + "teaching_rms": 107423244482.05092, + "traffic_rms": 1902228515095.4216 + }, + { + "batch_loss": 5133.7919921875, + "innovation_rms": 113633570608.38858, + "instruction_rms": 0.06194319150708626, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.127863128133468e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.494841777563535e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 225651105726464.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.506591796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.783935546875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1028018206146560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1177321537011712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 647765088534528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2687860863926272.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 671326339596288.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 225651105726464.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.60249328613281, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.7838592529297, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1028018206146560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2013013150411.5747, + "step": 218, + "teaching_rms": 113633570608.38858, + "traffic_rms": 2013013150411.5747 + }, + { + "batch_loss": 4222.9306640625, + "innovation_rms": 120249222996.49417, + "instruction_rms": 0.02295281445295624, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.297269694988288e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.9227980635221168e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202983191085056.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.53829193115234, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.41925048828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1048316557131776.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1245809353949184.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 684878303592448.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2748280316362752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 604193719058432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202983191085056.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.63418579101562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.41915893554688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1048316557131776.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2130106833154.245, + "step": 219, + "teaching_rms": 120249222996.49417, + "traffic_rms": 2130106833154.245 + }, + { + "batch_loss": 5399.26025390625, + "innovation_rms": 127245600808.06638, + "instruction_rms": 0.027910870899045294, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.491512167930266e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.833077685006445e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182580049608704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.66565704345703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.48797607421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1066574530215936.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1318230152970240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 724207386230784.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2802657823555584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 543774333730816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182580049608704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.76154327392578, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.48788452148438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1066574530215936.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2253884693810.317, + "step": 220, + "teaching_rms": 127245600808.06638, + "traffic_rms": 2253884693810.317 + }, + { + "batch_loss": 4019.516845703125, + "innovation_rms": 134703189332.91008, + "instruction_rms": 0.02429339344388411, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.972703811099689e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.982174382104088e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164215390208000.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.87337493896484, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95.54554748535156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1082996069236736.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1394817338703872.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 765871186247680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2851597633716224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 489396893646848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164215390208000.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.96925354003906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95.54545593261719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1082996069236736.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2384752285934.1587, + "step": 221, + "teaching_rms": 134703189332.91008, + "traffic_rms": 2384752285934.1587 + }, + { + "batch_loss": 8981.1806640625, + "innovation_rms": 142654102618.02798, + "instruction_rms": 0.050110595084456123, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.479342276504453e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.7928790030785783e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147685554257920.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.09632110595703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.54222106933594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1097764649828352.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1475817066463232.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 809996673613824.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2895643328643072.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 440457184149504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147685554257920.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.19219207763672, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.54212188720703, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1097764649828352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2523135671644.7007, + "step": 222, + "teaching_rms": 142654102618.02798, + "traffic_rms": 2523135671644.7007 + }, + { + "batch_loss": 437.7945251464844, + "innovation_rms": 150905238339.39783, + "instruction_rms": 0.027685445654326953, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.013008799304253e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.6417181681865682e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132807225507840.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.495361328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.38756561279297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111045359796224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1561489047552000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 856719676669952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2935284534607872.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 396411455668224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132807225507840.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.59122467041016, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.38746643066406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1111045359796224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2669484003029.703, + "step": 223, + "teaching_rms": 150905238339.39783, + "traffic_rms": 2669484003029.703 + }, + { + "batch_loss": 504.5787658691406, + "innovation_rms": 160207434822.39844, + "instruction_rms": 0.03568298499982993, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.57537463717331e+17, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.494839677375456e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119415391649792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.02058410644531, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.1495361328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1122986912382976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1652107488788480.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 906184680800256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2970961485758464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 356770316812288.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119415391649792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.11643981933594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.14944458007812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1122986912382976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2824271831369.766, + "step": 224, + "teaching_rms": 160207434822.39844, + "traffic_rms": 2824271831369.766 + }, + { + "batch_loss": 722.9320678710938, + "innovation_rms": 170425458468.20963, + "instruction_rms": 0.034962919146080525, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0168203819079434e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.354890942790506e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142049634418688.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.64103698730469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.7344970703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1133723055554560.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1747962032029696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 958545633738752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3003070929698816.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 321093264998400.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 142049634418688.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.73688507080078, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.73440551757812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1133723055554560.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2987964473280.5166, + "step": 225, + "teaching_rms": 170425458468.20963, + "traffic_rms": 2987964473280.5166 + }, + { + "batch_loss": 483.4433288574219, + "innovation_rms": 178677344578.51, + "instruction_rms": 0.08626987027111432, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0793359330036941e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2725378521036363e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114907219492864.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.6374740600586, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60.06037139892578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1143374249721856.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849358693695488.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1013966079787008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3031969348714496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288983921721344.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114907236270080.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.69635009765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 60.060279846191406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1143374249721856.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3161038370523.6924, + "step": 226, + "teaching_rms": 178677344578.51, + "traffic_rms": 3161038370523.6924 + }, + { + "batch_loss": 329.6751708984375, + "innovation_rms": 189573053088.75888, + "instruction_rms": 0.028499234172830346, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1452809295859548e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.5265927426690605e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 183398643531776.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.45726776123047, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.022979736328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1152048875700224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1956620670074880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1072619763793920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3057977791610880.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 260085519482880.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 183398626754560.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.5161361694336, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.022884368896484, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1152048875700224.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3344265676857.0166, + "step": 227, + "teaching_rms": 189573053088.75888, + "traffic_rms": 3344265676857.0166 + }, + { + "batch_loss": 728.8541870117188, + "innovation_rms": 201782132627.91666, + "instruction_rms": 0.06934573700830335, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2148630419134218e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.810631057737658e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191829144240128.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.89376068115234, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.61161422729492, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1159844509777920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2070089813721088.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1134690899591168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3081385363374080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273040231366656.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191829127462912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.95262145996094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.61151885986328, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1159844509777920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3538207111045.183, + "step": 228, + "teaching_rms": 201782132627.91666, + "traffic_rms": 3538207111045.183 + }, + { + "batch_loss": 504.798095703125, + "innovation_rms": 213789738023.4401, + "instruction_rms": 0.02890449492919065, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2883018974337434e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.1183472004810964e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 149452178325504.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 117.98440551757812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.741966247558594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1166849064566784.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2190127304540160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1200374908190720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3102452177960960.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311538338496512.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 149452195102720.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.04325866699219, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.74187088012695, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1166849064566784.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3743229824257.4224, + "step": 229, + "teaching_rms": 213789738023.4401, + "traffic_rms": 3743229824257.4224 + }, + { + "batch_loss": 522.3966064453125, + "innovation_rms": 228414633540.79172, + "instruction_rms": 0.041371843556968, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3658291495029965e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.582536643052633e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 313304140480512.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.76480865478516, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.12720489501953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1173141460090880.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2317115260403712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1269879155982336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3121412311089152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 382564481105920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 313304174034944.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.82365417480469, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.12724304199219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1173141460090880.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3960181298173.443, + "step": 230, + "teaching_rms": 228414633540.79172, + "traffic_rms": 3960181298173.443 + }, + { + "batch_loss": 380.7322998046875, + "innovation_rms": 243038728981.14697, + "instruction_rms": 0.029517817286910268, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4476895768973148e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.863599163037733e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 398285772685312.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.26596069335938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.730987548828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1178792831746048.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2451457542455296.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1343422820515840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3138476484591616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 462989018791936.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 398285806239744.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.32479858398438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.73102569580078, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1178792831746048.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4189358944512.7295, + "step": 231, + "teaching_rms": 243038728981.14697, + "traffic_rms": 4189358944512.7295 + }, + { + "batch_loss": 618.1134643554688, + "innovation_rms": 263059215970.91098, + "instruction_rms": 0.03120375148087112, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5341415648492257e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0095507792700665e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 507869346660352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.51576232910156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.506690979003906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1183867335606272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2593581499940864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1421238635331584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3153834213900288.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 613605267472384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 507869346660352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.57460021972656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.50675582885742, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1183867335606272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4432779867242.625, + "step": 232, + "teaching_rms": 263059215970.91098, + "traffic_rms": 4432779867242.625 + }, + { + "batch_loss": 1146.72509765625, + "innovation_rms": 268113363673.71375, + "instruction_rms": 0.04183147923708107, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.625458479437185e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0298868845229221e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 356096778698752.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.5393295288086, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.48398971557617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1188422551076864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2743938909732864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1503572755742720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3167656223965184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 606742178168832.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 356096812253184.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.59815979003906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.48405456542969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1188422551076864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4688524337977.418, + "step": 233, + "teaching_rms": 268113363673.71375, + "traffic_rms": 4688524337977.418 + }, + { + "batch_loss": 1227.9344482421875, + "innovation_rms": 282758778043.3844, + "instruction_rms": 0.04351118973531663, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7219288050245304e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1970917318090806e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303644792586240.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 129.35926818847656, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.061065673828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1192510286200832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2903007586942976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1590685966794752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3180096059867136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 587315571326976.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303644826140672.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 129.4180908203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.06117248535156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1192510286200832.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4960107034656.217, + "step": 234, + "teaching_rms": 282758778043.3844, + "traffic_rms": 4960107034656.217 + }, + { + "batch_loss": 1499.3389892578125, + "innovation_rms": 301037098821.89386, + "instruction_rms": 0.045798366477967194, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8238573812100628e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.899672531190115e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 490034964725760.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 130.99591064453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.00849151611328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1196177382965248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3071292995534848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1682854085918720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3191291965865984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 585664156401664.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 490034964725760.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 131.0547332763672, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.00828552246094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1196177382965248.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5247518581423.257, + "step": 235, + "teaching_rms": 301037098821.89386, + "traffic_rms": 5247518581423.257 + }, + { + "batch_loss": 1650.9434814453125, + "innovation_rms": 330279604295.3089, + "instruction_rms": 0.04357708424956132, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.931566502339674e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.107911603567951e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1262131571851264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.46759033203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 88.0943603515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1199465851518976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3249329858936832.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1780369573543936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3201368227577856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 694693243387904.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1262131571851264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.52639770507812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 88.0941162109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1199465851518976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5552105886200.731, + "step": 236, + "teaching_rms": 330279604295.3089, + "traffic_rms": 5552105886200.731 + }, + { + "batch_loss": 2181.42236328125, + "innovation_rms": 354909615161.81555, + "instruction_rms": 0.06057026004568144, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0453966047011144e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.1889293521429616e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 576356559618048.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 133.790771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.52371215820312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1202413407043584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3437684039090176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1883541533097984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3210436782587904.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 846138756300800.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 576356559618048.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 133.84957885742188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.52347564697266, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1202413407043584.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5876464591885.603, + "step": 237, + "teaching_rms": 354909615161.81555, + "traffic_rms": 5876464591885.603 + }, + { + "batch_loss": 5169.39208984375, + "innovation_rms": 375922686042.37506, + "instruction_rms": 0.06601821268294342, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1657076409135268e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.460491367659273e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 722845277618176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.9803009033203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128.4516143798828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1205054140841984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3636953878626304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1992697321619456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3218598562627584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 978022370902016.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 722845277618176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.03909301757812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128.4512939453125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1205054140841984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6217541260077.191, + "step": 238, + "teaching_rms": 375922686042.37506, + "traffic_rms": 6217541260077.191 + }, + { + "batch_loss": 2569.8837890625, + "innovation_rms": 441957946738.9089, + "instruction_rms": 0.05280712416688111, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2928800420001219e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.931539561664309e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1697090157674496.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.04953002929688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.60942077636719, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1207418788773888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3847772079915008.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2108182683975680.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3225944030445568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1459276610535424.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1697090157674496.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.10830688476562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.60910034179688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1207418788773888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6579853435309.282, + "step": 239, + "teaching_rms": 441957946738.9089, + "traffic_rms": 6579853435309.282 + }, + { + "batch_loss": 2554.37060546875, + "innovation_rms": 389638122850.30414, + "instruction_rms": 0.04396281244321669, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4273159543387587e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.365353448386688e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1527345265508352.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.01046752929688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.05029296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1209534865473536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4070808657854464.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2230364571435008.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3232555058855936.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1313348922638336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1527345265508352.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.06924438476562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.04997253417969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1209534865473536.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6952933073446.371, + "step": 240, + "teaching_rms": 389638122850.30414, + "traffic_rms": 6952933073446.371 + }, + { + "batch_loss": 2132.98779296875, + "innovation_rms": 412174047356.83606, + "instruction_rms": 0.031113961464047584, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.569440751490433e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.864965086137125e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1374559588581376.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.8700714111328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.64603424072266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1211427335438336.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4306771745177600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2359630873231360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3238504930738176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1182014057218048.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1374559588581376.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.92884826660156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.64571380615234, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1211427335438336.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7355804432209.473, + "step": 241, + "teaching_rms": 412174047356.83606, + "traffic_rms": 7355804432209.473 + }, + { + "batch_loss": 2156.63623046875, + "innovation_rms": 436030217056.6525, + "instruction_rms": 0.04238897384871468, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7197035839550915e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.424943631963274e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1237038762295296.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.69134521484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.28126525878906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1213118344593408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4556411082113024.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2496391490306048.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3243859949649920.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063812597809152.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1237038762295296.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.7501220703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.28094482421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1213118344593408.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7781960867127.833, + "step": 242, + "teaching_rms": 436030217056.6525, + "traffic_rms": 7781960867127.833 + }, + { + "batch_loss": 2084.15771484375, + "innovation_rms": 461315900320.8443, + "instruction_rms": 0.03476941616945349, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.878579990511747e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.0408386914239044e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1113257603497984.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.40365600585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.85212707519531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1214628159815680.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4820519358562304.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2641080885444608.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3248679439826944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 957431291052032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1113257603497984.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.46243286132812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.85181427001953, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1214628159815680.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8232775611799.817, + "step": 243, + "teaching_rms": 461315900320.8443, + "traffic_rms": 8232775611799.817 + }, + { + "batch_loss": 2561.34326171875, + "innovation_rms": 488250583333.16675, + "instruction_rms": 0.03735896906196583, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0465721730963866e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.709194554899351e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1001843467485184.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.9750213623047, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.26515197753906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1215974900498432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5099935435325440.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2794158083276800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3253016819924992.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 861688148525056.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1001843467485184.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.03378295898438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.26483917236328, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1215974900498432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8709693693042.017, + "step": 244, + "teaching_rms": 488250583333.16675, + "traffic_rms": 8709693693042.017 + }, + { + "batch_loss": 2595.93701171875, + "innovation_rms": 516772312003.9683, + "instruction_rms": 0.03582277378973062, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.2242117455810396e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.427485515224462e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 901560745852928.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.4071807861328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.436187744140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1226346441211904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5395546223149056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2956109488848896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3256920676761600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 775519293407232.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 901560745852928.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.46592712402344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.435874938964844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1226346575429632.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9214269444983.484, + "step": 245, + "teaching_rms": 516772312003.9683, + "traffic_rms": 9214269444983.484 + }, + { + "batch_loss": 2918.273681640625, + "innovation_rms": 547396302966.3827, + "instruction_rms": 0.03564051194214327, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.412060283529593e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.1938620899805445e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 811297176289280.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.7062225341797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.03957748413086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1299348402995200.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5708291246129152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3127448619188224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3260433960009728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 697967350644736.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 811297176289280.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.7649688720703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.03960037231445, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1299348537212928.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9748117632434.129, + "step": 246, + "teaching_rms": 547396302966.3827, + "traffic_rms": 9748117632434.129 + }, + { + "batch_loss": 21204.0625, + "innovation_rms": 580024770487.2722, + "instruction_rms": 0.17278802893420053, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.610712347854766e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.0068836264468807e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 730051897196544.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.7466583251953, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287.501708984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1365037243432960.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6039163178582016.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3308719861399552.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3263595861245952.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 628170575314944.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 730051897196544.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.80540466308594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287.50146484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1365037377650688.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10312903559353.588, + "step": 247, + "teaching_rms": 580024770487.2722, + "traffic_rms": 10312903559353.588 + }, + { + "batch_loss": 4897.41015625, + "innovation_rms": 614208847905.2571, + "instruction_rms": 0.09880415186168538, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8207963094518333e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.866418902979306e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 656923837857792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.1145477294922, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 258.7536926269531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1424143509618688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6389213213753344.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3500499277971456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3266441813950464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 565353524494336.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 656923837857792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.1732940673828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 258.75347900390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1424143643836416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10910237899212.947, + "step": 248, + "teaching_rms": 614208847905.2571, + "traffic_rms": 10910237899212.947 + }, + { + "batch_loss": 2908.8134765625, + "innovation_rms": 668402854166.4982, + "instruction_rms": 0.33658734638030324, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0429765482218783e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.376725082712038e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 567128017076224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.54425048828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.87789916992188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1480856271060992.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6759552674430976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3703395143647232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3269002956636160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 614256424779776.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 567128017076224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.6029815673828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.87770080566406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1480856405278720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 11548990224312.904, + "step": 249, + "teaching_rms": 668402854166.4982, + "traffic_rms": 11548990224312.904 + }, + { + "batch_loss": 298.2196960449219, + "innovation_rms": 724134448684.6107, + "instruction_rms": 0.0344876366249945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.2779562018508636e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.151839253866337e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 715783680294912.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.0295867919922, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 209.5873565673828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1534326734848000.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7151357844783104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3918051971956736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3271308011896832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 755730634244096.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 715783680294912.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.0883026123047, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 209.587158203125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1534326869065728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12222362402883.328, + "step": 250, + "teaching_rms": 724134448684.6107, + "traffic_rms": 12222362402883.328 + }, + { + "batch_loss": 321.4407653808594, + "innovation_rms": 785962666934.6815, + "instruction_rms": 0.03464644147026359, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.526478265321259e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.390669609657993e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 739252723777536.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.56500244140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.6174774169922, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1581303208083456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7565873191583744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4145151320522752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3273382749536256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1029317232427008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 739252790886400.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.6237030029297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.6172637939453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1581303342301184.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12936325536942.418, + "step": 251, + "teaching_rms": 785962666934.6815, + "traffic_rms": 12936325536942.418 + }, + { + "batch_loss": 340.29107666015625, + "innovation_rms": 834795352252.3158, + "instruction_rms": 0.050176592528576716, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.789329164324831e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.057502618805543e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 745664841515008.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.14549255371094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 169.74899291992188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1623649706573824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8004414585438208.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4385414206980096.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3275249986568192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1305673547644928.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 745664841515008.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.2041778564453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 169.74879455566406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1623649840791552.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13687201858251.129, + "step": 252, + "teaching_rms": 834795352252.3158, + "traffic_rms": 13687201858251.129 + }, + { + "batch_loss": 245.68153381347656, + "innovation_rms": 881671936596.6129, + "instruction_rms": 0.04407860298233351, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.067340129652179e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2235015577509326e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 741324844171264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.76654052734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 152.7656707763672, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1662443193368576.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8468375206363136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4639604061765632.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3276930392522752.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1554929558749184.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 741324844171264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.82522583007812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 152.76547241210938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1662443327586304.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14481245966723.592, + "step": 253, + "teaching_rms": 881671936596.6129, + "traffic_rms": 14481245966723.592 + }, + { + "batch_loss": 5869.23291015625, + "innovation_rms": 865052459366.6661, + "instruction_rms": 0.053607373983479346, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.361390770605523e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.0161128618002882e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 667152302473216.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.4241180419922, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.47915649414062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1697340742172672.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8959228228141056.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4908527533424640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3278442757881856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1399436576030720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 667152302473216.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.4827880859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.4789581298828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1697340876390400.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15297463870092.477, + "step": 254, + "teaching_rms": 865052459366.6661, + "traffic_rms": 15297463870092.477 + }, + { + "batch_loss": 5168.8603515625, + "innovation_rms": 915610625876.337, + "instruction_rms": 0.03845306475852671, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.672411823777776e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.3813019237968722e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 613272004853760.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.11456298828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.71991729736328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1728731450179584.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9478532187029504.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5193039052013568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3279803994079232.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1290965566357504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 613272004853760.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.1732177734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.71971130371094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1728731584397312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16184512113586.46, + "step": 255, + "teaching_rms": 915610625876.337, + "traffic_rms": 16184512113586.46 + }, + { + "batch_loss": 4666.76123046875, + "innovation_rms": 975491272014.9169, + "instruction_rms": 0.04213719940464446, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.001386802319983e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.5054605010669798e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 848555749146624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.8345947265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.33536529541016, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1756965894094848.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0027936887341056e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5494042171277312.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3281029133500416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1454685290496000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 848555749146624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 136.9600372314453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.33515930175781, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1756966028312576.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17123747523819.61, + "step": 256, + "teaching_rms": 975491272014.9169, + "traffic_rms": 17123747523819.61 + }, + { + "batch_loss": 5441.4267578125, + "innovation_rms": 1037794283365.817, + "instruction_rms": 0.04203585644471487, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.349357493499462e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.975847231721957e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1487643795783680.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.74742126464844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.92108154296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1782359351361536.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.06091866226688e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5812491984568320.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3282131597918208.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1700336481861632.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1487643795783680.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.9788360595703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.921142578125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1782359485579264.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 18116501396501.7, + "step": 257, + "teaching_rms": 1037794283365.817, + "traffic_rms": 18116501396501.7 + }, + { + "batch_loss": 4753.66552734375, + "innovation_rms": 1092392105853.997, + "instruction_rms": 0.04110020310103909, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.717423958699803e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.901930594680206e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1138865473585152.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.7628936767578, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.93485260009766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1805195558256640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1224127155208192e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6149400493555712.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3283124003799040.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845022387339264.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1138865473585152.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.99427795410156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.9349136352539, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1805195692474368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 19165386446556.77, + "step": 258, + "teaching_rms": 1092392105853.997, + "traffic_rms": 19165386446556.77 + }, + { + "batch_loss": 3727.5400390625, + "innovation_rms": 1143061174862.916, + "instruction_rms": 0.04014055173912982, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.10675058073482e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.2250040342694966e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1024923413774336.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.87525939941406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.4556884765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1825730065334272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1874710547595264e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6505837145096192.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3284017088561152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1660520121761792.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1024923413774336.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 164.1066131591797, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.4556884765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1825730199552000.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 20272981271765.312, + "step": 259, + "teaching_rms": 1143061174862.916, + "traffic_rms": 20272981271765.312 + }, + { + "batch_loss": 2941.483154296875, + "innovation_rms": 1220791820458.9211, + "instruction_rms": 0.037827210713308614, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.518569362383438e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.6165821038367495e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1097275963080704.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.17474365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.00186157226562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844192921780224.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2563003752841216e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6882934199943168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3284820784316416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862232589729792.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1097275963080704.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.4060821533203, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.00186157226562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844193055997952.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 21452762140540.64, + "step": 260, + "teaching_rms": 1220791820458.9211, + "traffic_rms": 21452762140540.64 + }, + { + "batch_loss": 3566.10791015625, + "innovation_rms": 1328694644809.2307, + "instruction_rms": 0.041326456951859915, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.954183774680383e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.363146718048324e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2456399976071168.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.74256896972656, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.2756576538086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860791091331072.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3291192909299712e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7281888880230400.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3285544217870336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2403291363278848.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2456399976071168.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.97389221191406, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.27565002441406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860791225548800.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22699772895742.145, + "step": 261, + "teaching_rms": 1328694644809.2307, + "traffic_rms": 22699772895742.145 + }, + { + "batch_loss": 2256.53369140625, + "innovation_rms": 1366328469697.2468, + "instruction_rms": 0.04422256846969644, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.414970955939447e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.054349541491301e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2210749858447360.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 183.65184020996094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.55166625976562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1875710733975552.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4061589783117824e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7703967664439296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3286195173851136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2974958489698304.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2210749858447360.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 183.8831329345703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.5517578125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1875710868193280.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24006873392080.133, + "step": 262, + "teaching_rms": 1366328469697.2468, + "traffic_rms": 24006873392080.133 + }, + { + "batch_loss": 3081.048095703125, + "innovation_rms": 1422312917084.7463, + "instruction_rms": 0.04715586145688667, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.9023905078465e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2594652052887827e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1989642660347904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.9683380126953, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.36767578125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889119756091392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4876641431912448e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8150512192978944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3286781168451584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2677462479667200.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1989642660347904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 189.19961547851562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.36763763427734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889119890309120.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 25393879570470.746, + "step": 263, + "teaching_rms": 1422312917084.7463, + "traffic_rms": 25393879570470.746 + }, + { + "batch_loss": 3295.471435546875, + "innovation_rms": 1506332834651.2007, + "instruction_rms": 0.050274468543519094, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.417983395947872e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.708308510324778e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1790626223882240.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.7512969970703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.86746215820312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1901168884187136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5738935573479424e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8622939805057024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3287308375687168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2409716097482752.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1790626223882240.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.98255920410156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.86753845214844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1901169018404864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 26865316902897.004, + "step": 264, + "teaching_rms": 1506332834651.2007, + "traffic_rms": 26865316902897.004 + }, + { + "batch_loss": 3919.7802734375, + "innovation_rms": 1592054436609.4148, + "instruction_rms": 0.0401116202438701, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.963380745743368e+18, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.2448779526744038e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1611493606948864.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.0540313720703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95.89002227783203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1911994080821248.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6651210249469952e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9122751054872576.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3287782969573376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2168744407203840.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1611493606948864.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.28526306152344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95.89009857177734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1911994215038976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28423281847483.887, + "step": 265, + "teaching_rms": 1592054436609.4148, + "traffic_rms": 28423281847483.887 + }, + { + "batch_loss": 4255.37890625, + "innovation_rms": 1681038175409.8254, + "instruction_rms": 0.043773556454711426, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0540308790488596e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.8249789762211066e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1450258118737920.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.92449951171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.98895263671875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1921717618343936.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7616363489067008e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9651532395970560.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3307847983038464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1951869899374592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1450258118737920.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 202.1557159423828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 108.9890365600586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1921717752561664.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30068390248642.06, + "step": 266, + "teaching_rms": 1681038175409.8254, + "traffic_rms": 30068390248642.06 + }, + { + "batch_loss": 4280.943359375, + "innovation_rms": 1778532893743.3037, + "instruction_rms": 0.0694535840341429, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1150593818997293e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.3836065757894866e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1305131576459264.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 205.40591430664062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.99178314208984, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1930449689509888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863746082517811e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0210963697434624e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3343419707490304.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1756682828906496.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1305131576459264.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 205.63710021972656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.99176025390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1930449823727616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31810763959351.887, + "step": 267, + "teaching_rms": 1778532893743.3037, + "traffic_rms": 31810763959351.887 + }, + { + "batch_loss": 3773.90478515625, + "innovation_rms": 1881960694290.763, + "instruction_rms": 0.06603741998927594, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1796166573687833e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.8602757747725846e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1174504709554176.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.53712463378906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 161.5145721435547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1938289212784640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.971774388436992e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0802820928241664e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3375434393714688.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1581014505750528.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1174504709554176.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.76829528808594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 161.5144805908203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1938289347002368.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33654118720508.277, + "step": 268, + "teaching_rms": 1881960694290.763, + "traffic_rms": 33654118720508.277 + }, + { + "batch_loss": 4362.10205078125, + "innovation_rms": 1991218076878.7905, + "instruction_rms": 0.052747455913279416, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2479068847652995e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.678735576327229e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1056928771866624.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 211.35313415527344, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.68617248535156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1945325442957312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.086064327177011e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1428984210325504e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432086622961664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1422913001488384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1056928771866624.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 211.5842742919922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.6860809326172, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1945325577175040.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35604391733989.77, + "step": 269, + "teaching_rms": 1991218076878.7905, + "traffic_rms": 35604391733989.77 + }, + { + "batch_loss": 4237.091796875, + "innovation_rms": 2106921725849.5671, + "instruction_rms": 0.04504228180864075, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3201462280752988e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.741352584158302e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 951099905277952.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.8854217529297, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202.38438415527344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1951638508011520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.206978823474381e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2091442113544192e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3516657750245376.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1280621674496000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 951099905277952.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.11654663085938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 202.3843994140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1951638642229248.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 37667662562391.47, + "step": 270, + "teaching_rms": 2106921725849.5671, + "traffic_rms": 37667662562391.47 + }, + { + "batch_loss": 3902.328369140625, + "innovation_rms": 2230181547134.968, + "instruction_rms": 0.04266928272130071, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3965632757661958e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1667200335140006e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 855844375756800.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.16233825683594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.21038818359375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1957300751302656.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.334901847405363e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.279229702438912e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3592771818487808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1152559439937536.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 855844375756800.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.39344787597656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.21041870117188, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1957300885520384.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 39850530490980.9, + "step": 271, + "teaching_rms": 2230181547134.968, + "traffic_rms": 39850530490980.9 + }, + { + "batch_loss": 2953.86083984375, + "innovation_rms": 2361738680280.517, + "instruction_rms": 0.04564061649478734, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4773997004937757e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.2698744766332804e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 770105889783808.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 218.20941162109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.66741943359375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1962377268428800.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.47023970287616e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3533776957145088e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3661274399375360.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1469537153187840.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 770105889783808.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 218.44049072265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.66744995117188, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1962377402646528.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 42159953085307.28, + "step": 272, + "teaching_rms": 2361738680280.517, + "traffic_rms": 42159953085307.28 + }, + { + "batch_loss": 9307.521484375, + "innovation_rms": 2502126042185.7896, + "instruction_rms": 0.03850703848600795, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5629114685649846e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.5207598815189526e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 692933581864960.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.04959106445312, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287.2489929199219, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1966926444101632.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.613422101364736e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4318235553890304e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3722926775861248.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2007436307202048.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 692933581864960.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.28065490722656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 287.2490539550781, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1966926578319360.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 44603146561093.04, + "step": 273, + "teaching_rms": 2502126042185.7896, + "traffic_rms": 44603146561093.04 + }, + { + "batch_loss": 11520.9228515625, + "innovation_rms": 2639672632143.8535, + "instruction_rms": 0.04803671775051471, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6533690598402556e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.1210767919050136e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 623471545548800.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 221.70355224609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 378.8324890136719, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1971001025888256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.764903665159373e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5148162821914624e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3778413995229184.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1806692622794752.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 623471545548800.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 221.93460083007812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 378.832275390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1971001160105984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 47187663274559.82, + "step": 274, + "teaching_rms": 2639672632143.8535, + "traffic_rms": 47187663274559.82 + }, + { + "batch_loss": 12340.5478515625, + "innovation_rms": 2792704002921.936, + "instruction_rms": 0.04898211836784421, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7490585672451359e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.000623678657497e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1098168578080768.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 223.1898956298828, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 289.8707580566406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1974648526864384.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.925165645345587e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6026195871137792e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3828352385286144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1626023347093504.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1098168578080768.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 223.42092895507812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 289.8705749511719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1974648661082112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 49922781101074.23, + "step": 275, + "teaching_rms": 2792704002921.936, + "traffic_rms": 49922781101074.23 + }, + { + "batch_loss": 4136.6396484375, + "innovation_rms": 2950870958470.5415, + "instruction_rms": 0.05705316119403133, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8502830161842405e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.198031079672323e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 880700156805120.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 224.52537536621094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 251.7461395263672, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1977911494049792.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0947167807995904e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6955121061593088e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3873297070555136.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1463409711251456.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 880700156805120.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 224.7563934326172, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 251.7459716796875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1977911628267520.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 52815935288301.016, + "step": 276, + "teaching_rms": 2950870958470.5415, + "traffic_rms": 52815935288301.016 + }, + { + "batch_loss": 6130.134765625, + "innovation_rms": 3120695197890.9927, + "instruction_rms": 0.07721074600336612, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.957362474492415e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.681355046441811e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 792594070110208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 225.72506713867188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.94297790527344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1980828313714688.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.274095660420301e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7937890109554688e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3913747072548864.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1316896565624832.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 792594070110208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 225.9560546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.94281005859375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1980828447932416.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 55875403246669.73, + "step": 277, + "teaching_rms": 3120695197890.9927, + "traffic_rms": 55875403246669.73 + }, + { + "batch_loss": 5925.46435546875, + "innovation_rms": 3301384751502.42, + "instruction_rms": 0.05773297983484053, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.070635921604503e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.456160506298812e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 713290720215040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 226.8025360107422, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.84036254882812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1983433748250624.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.463871796871168e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.897762223502131e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3950152289091584.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1186249163407360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 713290720215040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 227.0334930419922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.84017944335938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1983433882468352.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 59113417173647.59, + "step": 278, + "teaching_rms": 3301384751502.42, + "traffic_rms": 59113417173647.59 + }, + { + "batch_loss": 4925.89111328125, + "innovation_rms": 3492594946292.454, + "instruction_rms": 0.061874779358629804, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1904617983111594e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.309046386686903e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 641910578348032.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 227.76998901367188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164.2733917236328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1985758801952768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.66464820356055e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.007762134158541e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3982916983980032.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1071426165538816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 641910578348032.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 228.00091552734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 164.27320861816406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1985758936170496.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 62539179548422.61, + "step": 279, + "teaching_rms": 3492594946292.454, + "traffic_rms": 62539179548422.61 + }, + { + "batch_loss": 5087.12255859375, + "innovation_rms": 3694966533975.605, + "instruction_rms": 0.058138882463565744, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.317219216221641e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.8385455279302816e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 577661961240576.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 228.63841247558594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.07345581054688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2011607391535104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.877061824138445e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1241379237658624e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4012405155692544.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 974046304927744.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 577661961240576.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 228.86932373046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.07327270507812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2011607391535104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 66163573139921.625, + "step": 280, + "teaching_rms": 3694966533975.605, + "traffic_rms": 66163573139921.625 + }, + { + "batch_loss": 4793.279296875, + "innovation_rms": 3910203463254.2446, + "instruction_rms": 0.06644013756981526, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4513094970800865e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.6595603603547583e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1002198272049152.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 229.41770935058594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.49224853515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2041328162570240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.101787827463782e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.247259174259917e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4038944563920896.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 939174190383104.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1002198272049152.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 229.64859008789062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119.4920654296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2041328162570240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 69997986801129.58, + "step": 281, + "teaching_rms": 3910203463254.2446, + "traffic_rms": 69997986801129.58 + }, + { + "batch_loss": 7251.33251953125, + "innovation_rms": 4141877216085.3228, + "instruction_rms": 0.06586902715611613, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.593156612570166e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.0928088788226286e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1351796396130304.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.1167755126953, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.64254760742188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2068056415141888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.339539607604429e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.377516942413005e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4062829950795776.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1148094150344704.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1351796396130304.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.34764099121094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.64252471923828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2068056415141888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 74055051058714.75, + "step": 282, + "teaching_rms": 4141877216085.3228, + "traffic_rms": 74055051058714.75 + }, + { + "batch_loss": 6867.1357421875, + "innovation_rms": 4383884954252.7593, + "instruction_rms": 0.07750196374343185, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7432096032406634e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.917717259280142e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1071741778526208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.7432098388672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.81202697753906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2092091186348032.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.591072219811021e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.515324833575731e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4084326798983168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1460239488516096.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1071741778526208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.97406005859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.81201171875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2092091186348032.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 78346425662076.48, + "step": 283, + "teaching_rms": 4383884954252.7593, + "traffic_rms": 78346425662076.48 + }, + { + "batch_loss": 4188.9169921875, + "innovation_rms": 4627961850817.755, + "instruction_rms": 0.06692252454703405, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9019432382124524e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.2812724780429e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 964534831415296.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 231.3046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.4003677368164, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2113701582733312.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.857184098503885e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6611205049155584e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4103674016038912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1314215566508032.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 964534831415296.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 231.5355224609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.40042114257812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2113701582733312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 82885556284329.88, + "step": 284, + "teaching_rms": 4627961850817.755, + "traffic_rms": 82885556284329.88 + }, + { + "batch_loss": 5519.9052734375, + "innovation_rms": 4896111051324.328, + "instruction_rms": 0.06299985502901781, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.069859774397101e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.733095916456803e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868038928760832.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 231.80770874023438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.7838134765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2133129867296768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.138720922743603e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.815366953906995e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4121086350327808.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1182793996435456.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868038928760832.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.0385284423828, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.78392028808594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2133129867296768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 87689099764679.53, + "step": 285, + "teaching_rms": 4896111051324.328, + "traffic_rms": 87689099764679.53 + }, + { + "batch_loss": 6508.86328125, + "innovation_rms": 5179739573142.603, + "instruction_rms": 0.06948405598316505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.247490715715476e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.270244991048471e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 781183952617472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.25811767578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 212.4644317626953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2150593875410944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.436576475224474e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9785538068217856e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4136757612249088.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1064514556526592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 781183952617472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.48890686035156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 212.46408081054688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2150593875410944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 92771064346617.97, + "step": 286, + "teaching_rms": 5179739573142.603, + "traffic_rms": 92771064346617.97 + }, + { + "batch_loss": 4617.30810546875, + "innovation_rms": 5479743847224.822, + "instruction_rms": 0.09343512060077794, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.435397912609371e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.8800159611347565e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 703006689460224.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.66116333007812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.86660766601562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2166289967611904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.751696507745075e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.151199466212557e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4150861747978240.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 958063053897728.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 703006689460224.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.8919219970703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.86685180664062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2166289967611904.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 98147582008477.05, + "step": 287, + "teaching_rms": 5479743847224.822, + "traffic_rms": 98147582008477.05 + }, + { + "batch_loss": 6548.60888671875, + "innovation_rms": 5797234568491.708, + "instruction_rms": 0.10108139388799765, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6341759809670873e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.571473952138337e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 632640092766208.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.02157592773438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 174.3894500732422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2180394908647424.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.085081747685376e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.333852399403008e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4163555523821568.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 862256694820864.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 632640092766208.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.25230407714844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 174.38970947265625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2180394908647424.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 103835738651540.52, + "step": 288, + "teaching_rms": 5797234568491.708, + "traffic_rms": 103835738651540.52 + }, + { + "batch_loss": 8983.3369140625, + "innovation_rms": 6133257797294.561, + "instruction_rms": 0.08698454716064807, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.844454281244364e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.340592254596427e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 569303887773696.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.34361267089844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.33346557617188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2193067478089728.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.43779090448384e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.527092426978099e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4174979868393472.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 916805799378944.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 569303887773696.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.57432556152344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.33331298828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2193067478089728.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 109853578995616.58, + "step": 289, + "teaching_rms": 6133257797294.561, + "traffic_rms": 109853578995616.58 + }, + { + "batch_loss": 5262.50830078125, + "innovation_rms": 6488232171671.098, + "instruction_rms": 0.07274029918685977, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.06689779807368e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.1814603237914584e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 842106989969408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.63111877441406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 152.31900024414062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2204450886254592.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.810944105611264e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.731532870267699e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4185261751664640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 959944518008832.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 842106989969408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.86180114746094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 152.31884765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2204450886254592.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 116220211271826.02, + "step": 290, + "teaching_rms": 6488232171671.098, + "traffic_rms": 116220211271826.02 + }, + { + "batch_loss": 4949.4755859375, + "innovation_rms": 6863987970783.386, + "instruction_rms": 0.06872794160857093, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.302210438799137e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.078826738439435e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 742917538840576.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.8875274658203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 129.29208374023438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2214673847943168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.205726332544614e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.947823557823693e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4194515527139328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 963082260054016.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 742917538840576.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.11819458007812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 129.29193115234375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2214673847943168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 122955920866821.17, + "step": 291, + "teaching_rms": 6863987970783.386, + "traffic_rms": 122955920866821.17 + }, + { + "batch_loss": 4307.0634765625, + "innovation_rms": 7261158456802.112, + "instruction_rms": 0.07844153106484976, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.551136792695066e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.0219653670699693e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 668600981520384.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.115966796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 106.24183654785156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2223852461490176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.623391715734323e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.17665125491671e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4202843737161728.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 866773993783296.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 668600981520384.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.3466033935547, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 106.24186706542969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2223852461490176.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 130081884337326.25, + "step": 292, + "teaching_rms": 7261158456802.112, + "traffic_rms": 130081884337326.25 + }, + { + "batch_loss": 4984.24462890625, + "innovation_rms": 7681815867493.776, + "instruction_rms": 0.08536678291671383, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.814464989696256e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.0274670820895633e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 601709483130880.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.3192138671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.01643371582031, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2232090879852544.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.065265682087936e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.418742240516506e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4210339260399616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 780096587694080.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 601709483130880.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.54983520507812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.01646423339844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2232090879852544.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 137620993607946.08, + "step": 293, + "teaching_rms": 7681815867493.776, + "traffic_rms": 137620993607946.08 + }, + { + "batch_loss": 5972.001953125, + "innovation_rms": 8126949876639.071, + "instruction_rms": 0.08078036707116076, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0930275800072585e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.095314861822677e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541501088071680.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.4998016357422, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 92.71280670166016, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2239483189657600.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.532752256414515e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.674865743265792e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4217085311844352.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 702086895370240.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541501088071680.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.7303924560547, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 92.71269226074219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2239483189657600.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 145597065815501.06, + "step": 294, + "teaching_rms": 8126949876639.071, + "traffic_rms": 145597065815501.06 + }, + { + "batch_loss": 5917.13916015625, + "innovation_rms": 8597614155861.048, + "instruction_rms": 0.09022546144238139, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.387706371953551e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.226595927108813e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487308096700416.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.65997314453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.6408920288086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2246113813856256.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.027335779411558e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.945834800473702e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4223156784988160.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 631878205833216.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487308096700416.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.8905487060547, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.64085388183594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2246113813856256.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 154035441851825.44, + "step": 295, + "teaching_rms": 8597614155861.048, + "traffic_rms": 154035441851825.44 + }, + { + "batch_loss": 4863.3212890625, + "innovation_rms": 9095694349849.594, + "instruction_rms": 0.08377141670536696, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.699434191200138e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.42292393278918e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438529515585536.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.80178833007812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.73729705810547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2252058853900288.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.550586491122483e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.232509694089626e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4228621057130496.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568690378539008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438529515585536.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.03233337402344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85.73726654052734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2252058853900288.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 162962938364431.62, + "step": 296, + "teaching_rms": 9095694349849.594, + "traffic_rms": 162962938364431.62 + }, + { + "batch_loss": 5005.1083984375, + "innovation_rms": 9622842711008.607, + "instruction_rms": 0.09514473425194656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.029197079774809e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.686159812227833e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 711618904195072.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.9270782470703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.97499084472656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2257387029266432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0104166543890842e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.535801386677043e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4233538794684416.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 564191937167360.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 711618904195072.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.1575927734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.97511291503906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2257387029266432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 172407911131074.72, + "step": 297, + "teaching_rms": 9622842711008.607, + "traffic_rms": 172407911131074.72 + }, + { + "batch_loss": 7450.94873046875, + "innovation_rms": 10180350745370.705, + "instruction_rms": 0.1395302511824075, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.378038254309999e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.018445201692736e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 481081501417472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.03749084472656, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 161.8189697265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2262159811674112.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0689833438334157e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.856672380406989e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4237964758482944.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 631360192512000.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 481081501417472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.26797485351562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 161.81907653808594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2262159811674112.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 182400360233017.16, + "step": 298, + "teaching_rms": 10180350745370.705, + "traffic_rms": 182400360233017.16 + }, + { + "batch_loss": 10949.6611328125, + "innovation_rms": 10769849058130.537, + "instruction_rms": 0.3904984325759337, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.747062064284644e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.422242363922028e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 432952970313728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.13450622558594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.92202758789062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2266432767262720.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1309447754285056e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.196142730012262e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4241948072214528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 568224173260800.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 432952970313728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.36497497558594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.922119140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2266432767262720.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 192971858443828.84, + "step": 299, + "teaching_rms": 10769849058130.537, + "traffic_rms": 192971858443828.84 + }, + { + "batch_loss": 10233.9130859375, + "innovation_rms": 11393839336825.148, + "instruction_rms": 0.4271752046186149, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.137436191047444e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.900375040145319e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 389632957087744.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.21946716308594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.5543975830078, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2270255556591616.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1964976586765107e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.555289183793971e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4245533027729408.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 511401755934720.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 389632957087744.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.44992065429688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.55447387695312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2270255556591616.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 204156197537960.97, + "step": 300, + "teaching_rms": 11393839336825.148, + "traffic_rms": 204156197537960.97 + }, + { + "batch_loss": 14759.4580078125, + "innovation_rms": 12053997234866.098, + "instruction_rms": 0.30823600332106155, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.550395606058716e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.456072087470074e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 350641062936576.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.2935791015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 292.02783203125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2273673545252864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2658502135919411e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.935252914562662e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4248759621910528.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 460261580341248.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 350641062936576.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.52401733398438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 292.0276794433594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2273673545252864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 215988789590745.06, + "step": 301, + "teaching_rms": 12053997234866.098, + "traffic_rms": 215988789590745.06 + }, + { + "batch_loss": 8341.2734375, + "innovation_rms": 12752420747534.164, + "instruction_rms": 0.14434698105967425, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.987247408741561e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.093010187812972e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 315544838340608.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.35792541503906, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.22254943847656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2276726998564864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3392226001983898e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.337240378631782e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4251663556673536.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 414235402174464.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 315544838340608.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.58834838867188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.22238159179688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2276726998564864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 228507257493452.28, + "step": 302, + "teaching_rms": 12752420747534.164, + "traffic_rms": 228507257493452.28 + }, + { + "batch_loss": 6243.0234375, + "innovation_rms": 13491322218229.486, + "instruction_rms": 0.11871400768262312, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.449373025505116e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.815373578506797e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 283955085443072.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.41348266601562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.84805297851562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2279452155314176.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4168478634213376e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.762528040281702e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4254277044273152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 372811851890688.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 283955085443072.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.64389038085938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.84780883789062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2279452155314176.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 241751306774367.22, + "step": 303, + "teaching_rms": 13491322218229.486, + "traffic_rms": 241751306774367.22 + }, + { + "batch_loss": 3765.705078125, + "innovation_rms": 14273050830108.809, + "instruction_rms": 0.0976563997954277, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.938236126228277e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 8.628192628901185e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 255521479196672.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.4611358642578, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.36039733886719, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2281882033061888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.498972620282921e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.212467096223744e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4256629075738624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 335530663346176.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 255521479196672.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.69151306152344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123.36013793945312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2281882033061888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 255763006671092.2, + "step": 304, + "teaching_rms": 14273050830108.809, + "traffic_rms": 255763006671092.2 + }, + { + "batch_loss": 5321.38330078125, + "innovation_rms": 15100084710550.006, + "instruction_rms": 0.12451703662585321, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.455383064064347e+19, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.536183890545813e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 229928675049472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.5016632080078, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.6339569091797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2284046159708160.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.585857575298007e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.688486911574016e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4258745957744640.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 301977573523456.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 229928675049472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.73202514648438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.6341552734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2284046159708160.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 270586870007297.03, + "step": 305, + "teaching_rms": 15100084710550.006, + "traffic_rms": 270586870007297.03 + }, + { + "batch_loss": 10308.517578125, + "innovation_rms": 15975055608083.238, + "instruction_rms": 0.1454982921423379, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0002449912315458e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0544507097524423e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206892852838400.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.53578186035156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.888427734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2285971110363136.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.677778637165691e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.192097596833792e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4260651312611328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271779809460224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 206892852838400.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.76612854003906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.88809204101562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2285971110363136.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 286269966032406.56, + "step": 306, + "teaching_rms": 15975055608083.238, + "traffic_rms": 286269966032406.56 + }, + { + "batch_loss": 8209.4501953125, + "innovation_rms": 16900743101343.867, + "instruction_rms": 0.1647749134101452, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0581169501306985e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1660404818036854e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186158545895424.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.5641326904297, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.61341857910156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2287680507346944.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7750276918634086e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.724899456817562e+16, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4262366078304256.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 244601826836480.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186158545895424.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.79446411132812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.61306762695312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2287680507346944.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 302862106996092.4, + "step": 307, + "teaching_rms": 16900743101343.867, + "traffic_rms": 302862106996092.4 + }, + { + "batch_loss": 11110.2998046875, + "innovation_rms": 17880088129586.156, + "instruction_rms": 0.2423673775750868, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1193371418387546e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2891330689059643e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167495805698048.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.58729553222656, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.3048553466797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2289196093931520.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8779134616403968e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.028858299065303e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4263909313740800.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220141635764224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167495805698048.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.81761169433594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.3045196533203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2289196093931520.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 320415934483171.5, + "step": 308, + "teaching_rms": 17880088129586.156, + "traffic_rms": 320415934483171.5 + }, + { + "batch_loss": 10554.9658203125, + "innovation_rms": 18916191603246.184, + "instruction_rms": 0.3331728278465627, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1840992563240632e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.4245330082908106e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150697668509696.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.6057891845703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 248.04736328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2290537197469696.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.986762879407227e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.088494005869609e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4265298198790144.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198127462121472.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150697668509696.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.83607482910156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 248.04721069335938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2290537197469696.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 338987223020914.9, + "step": 309, + "teaching_rms": 18916191603246.184, + "traffic_rms": 338987223020914.9 + }, + { + "batch_loss": 33250.4296875, + "innovation_rms": 20012353043885.836, + "instruction_rms": 0.5204673266377984, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2526082425540418e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.5732915586395855e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135577840123904.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.6200714111328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1466.260986328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2291721534701568.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.101921432333189e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1515863023537357e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4266548302708736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178314710876160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135577840123904.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 235.850341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1466.2608642578125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2291721534701568.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 358634980939383.0, + "step": 310, + "teaching_rms": 20012353043885.836, + "traffic_rms": 358634980939383.0 + }, + { + "batch_loss": 55425.4765625, + "innovation_rms": 21173714744528.016, + "instruction_rms": 4.708819575102043, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.325081012182618e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.7363636668985937e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.423261701622989e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.24603271484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1213.099365234375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7425742367358976.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2237550516319027e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2183356775910605e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4267673315704832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167098051461120.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.423261701622989e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.27027893066406, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1213.0992431640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7425742367358976.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 379423506617512.1, + "step": 311, + "teaching_rms": 21173714744528.016, + "traffic_rms": 379423506617512.1 + }, + { + "batch_loss": 135606.609375, + "innovation_rms": 22398845398511.688, + "instruction_rms": 17.011078261612703, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4017467913939504e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.9148792009416908e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.680861142627123e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 337.1561584472656, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1069.3834228515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4106603187863552e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3526504561586995e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.288953959368622e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4268685854244864.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150388246315008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.680861142627123e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 337.1803894042969, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1069.3836669921875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4106603187863552e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 401412992811815.25, + "step": 312, + "teaching_rms": 22398845398511.688, + "traffic_rms": 401412992811815.25 + }, + { + "batch_loss": 217366.265625, + "innovation_rms": 23696798408538.008, + "instruction_rms": 27.235647257366235, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4828481764335917e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.109919006217177e+34, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.012633809839718e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 411.2505187988281, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 998.9224853515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.01192374272e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4890170421962342e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3636655167779635e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4269597192617984.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207157848965120.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.012633809839718e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 411.2747497558594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 998.9225463867188, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.01192374272e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 424679065045877.94, + "step": 313, + "teaching_rms": 23696798408538.008, + "traffic_rms": 424679065045877.94 + }, + { + "batch_loss": 188530.828125, + "innovation_rms": 25070183642163.887, + "instruction_rms": 23.375434696054548, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.56864192525686e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.4504737771160673e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.411169166688256e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.6109924316406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 898.113037109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5530406593888256e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6332877424479437e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4427076897118618e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4270417262936064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 209446546440192.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.411169166688256e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.63519287109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 898.1130981445312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5530406593888256e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 449293688713936.1, + "step": 314, + "teaching_rms": 25070183642163.887, + "traffic_rms": 449293688713936.1 + }, + { + "batch_loss": 171999.21875, + "innovation_rms": 26523239750912.62, + "instruction_rms": 337.6873276458593, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6593993093725605e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.8438408302868652e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.869797128962048e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 501.0601806640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 776.4494018554688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0400203722850304e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.785920915823657e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5263312183610573e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4271155460440064.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122791965753344.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.869797128962048e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 501.0843505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 776.449462890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0400203722850304e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 475335019489173.5, + "step": 315, + "teaching_rms": 26523239750912.62, + "traffic_rms": 475335019489173.5 + }, + { + "batch_loss": 1858978.375, + "innovation_rms": 28060122077222.05, + "instruction_rms": 6.72779668870499, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.755407609178798e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.9358901195127913e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.382513332381286e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.6379699707031, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4910.025390625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.478271748472832e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.947401034634363e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6148018753018266e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4271819838193664.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 118043904573440.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.382513332381286e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.6621398925781, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4910.025390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.478271748472832e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 502885789608934.4, + "step": 316, + "teaching_rms": 28060122077222.05, + "traffic_rms": 502885789608934.4 + }, + { + "batch_loss": 2386115.25, + "innovation_rms": 29686265688301.06, + "instruction_rms": 4.754333845985912, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2048434543829346e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.083841589348681e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.943913849094144e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 588.8984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4996.26220703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.872663348130611e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.118241089773896e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.708400551395328e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4272417643954176.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 106239514116096.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.943913849094144e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 589.0419921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4996.26220703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.872663348130611e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 532033463713874.06, + "step": 317, + "teaching_rms": 29686265688301.06, + "traffic_rms": 532033463713874.06 + }, + { + "batch_loss": 3504192.0, + "innovation_rms": 31406664666014.242, + "instruction_rms": 7.218632319525251, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.079744032087086e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.12996765525007e+31, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.549135272882995e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 795.8189697265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6003.15625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.227576682145382e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.298983621511086e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8074244583784448e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4272955857043456.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95615560187904.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.549135272882995e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 795.9625244140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6003.15625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.227576682145382e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 562870651095675.94, + "step": 318, + "teaching_rms": 31406664666014.242, + "traffic_rms": 562870651095675.94 + }, + { + "batch_loss": 3114045.5, + "innovation_rms": 33226783092648.76, + "instruction_rms": 9.680324502092502, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.584558143818579e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.0151783386317791e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.193798884589568e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 978.2222290039062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6397.5986328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.546956463230157e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.49020243747668e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.912188159655936e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4273440114606080.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 86053998297088.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.193798884589568e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 978.3229370117188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6397.59765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.546956463230157e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 595495222867946.1, + "step": 319, + "teaching_rms": 33226783092648.76, + "traffic_rms": 595495222867946.1 + }, + { + "batch_loss": 9047358.0, + "innovation_rms": 35152395438558.938, + "instruction_rms": 6.057302981067109, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.97080549557908e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2862576127312132e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.873964266468147e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1302.487060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7812.919921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.834352911351808e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.692505017845023e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0230244292938957e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4273876053786624.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77448595111936.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.873964266468147e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1302.5877685546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7812.919921875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.834352911351808e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 630010780671251.0, + "step": 320, + "teaching_rms": 35152395438558.938, + "traffic_rms": 630010780671251.0 + }, + { + "batch_loss": 14358723.0, + "innovation_rms": 37189618591876.555, + "instruction_rms": 7.364356061825782, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.407405554079629e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.590078756730631e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.586084419777331e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1519.044921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8016.9287109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.09296148217856e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.906533546126213e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1402847674158285e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4274268237987840.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69703733084160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.586084419777331e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1519.1456298828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8016.92822265625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.09296148217856e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 666527001893868.8, + "step": 321, + "teaching_rms": 37189618591876.555, + "traffic_rms": 666527001893868.8 + }, + { + "batch_loss": 22830398.0, + "innovation_rms": 39344923244630.055, + "instruction_rms": 8.952043296190055, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.867406034769019e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.9319390901434525e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3269664658292736e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1597.7025146484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7236.75146484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.325658085811814e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.13296765794517e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2643418053843354e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4274621230612480.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62733357678592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3269664658292736e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1597.80322265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7236.75146484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.325658085811814e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 705159798936200.6, + "step": 322, + "teaching_rms": 39344923244630.055, + "traffic_rms": 705159798936200.6 + }, + { + "batch_loss": 33393036.0, + "innovation_rms": 41625146904043.15, + "instruction_rms": 10.521024765569852, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.3337579983078844e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.370148780384183e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.093737286251315e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2121.942138671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6099.30615234375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.5350317285376e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3725265026259354e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.395589821197189e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4274939058192384.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56460021071872.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.093737286251315e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2122.169189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6099.306640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.5350317285376e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 746031750903786.1, + "step": 323, + "teaching_rms": 41625146904043.15, + "traffic_rms": 746031750903786.1 + }, + { + "batch_loss": 37969232.0, + "innovation_rms": 44037538520823.53, + "instruction_rms": 13.07469317613504, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.7463992586345434e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.898761169703171e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.883810022241075e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2599.59814453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6087.2109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.723412859610726e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.6259711483733606e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5344454266821018e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4275224941953024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50814018125824.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.883810022241075e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2599.236083984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6087.2119140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.723412859610726e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 789272811690366.9, + "step": 324, + "teaching_rms": 44037538520823.53, + "traffic_rms": 789272811690366.9 + }, + { + "batch_loss": 42394456.0, + "innovation_rms": 46589754533804.9, + "instruction_rms": 7.038018719378422, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.345728797264545e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.446948099037097e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6948566941499392e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2401.444580078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5219.3544921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.89289871156183e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.8941059566626406e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6813494572823347e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4275482371555328.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45732614635520.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6948566941499392e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2401.08251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5219.3544921875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.89289871156183e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 835020264127645.8, + "step": 325, + "teaching_rms": 46589754533804.9, + "traffic_rms": 835020264127645.8 + }, + { + "batch_loss": 24804564.0, + "innovation_rms": 49289904201074.72, + "instruction_rms": 5.501698382508509, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.034194072300673e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.1448220956462634e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5247816800600064e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1933.1585693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14425.732421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.045376922517504e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.177782705407918e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8367681746475418e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4275714031353856.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41159351074816.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5247816800600064e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1933.2532958984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14425.734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.045376922517504e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 883419260760083.4, + "step": 326, + "teaching_rms": 49289904201074.72, + "traffic_rms": 883419260760083.4 + }, + { + "batch_loss": 17335944.0, + "innovation_rms": 52146554217970.73, + "instruction_rms": 5.527316519945181, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.755193632519902e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 5.037545533301375e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3716989524574208e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2369.050048828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12397.9130859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.182547011036774e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.477902306949202e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.001195671815455e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4275922605703168.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40712045330432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3716989524574208e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2369.144775390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12397.9140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.182547011036774e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 934623635151290.4, + "step": 327, + "teaching_rms": 52146554217970.73, + "traffic_rms": 934623635151290.4 + }, + { + "batch_loss": 13650972.0, + "innovation_rms": 55168778631622.37, + "instruction_rms": 5.246720502061326, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.463205233626259e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 6.13601847404834e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2339107107700736e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2774.220947265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12748.171875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.305937985477018e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.79541790042882e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1751538732118835e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276110242086912.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36640839958528.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2339107107700736e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2774.315673828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12748.171875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.305937985477018e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 988796010991783.4, + "step": 328, + "teaching_rms": 55168778631622.37, + "traffic_rms": 988796010991783.4 + }, + { + "batch_loss": 14858838.0, + "innovation_rms": 58366182933843.71, + "instruction_rms": 7.6816041208172265, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.094897709547289e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.548716187542064e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1098890203824128e+16, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3201.099609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10983.6708984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.416926812353331e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.131337256973107e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.359194939832402e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276279087988736.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32976754704384.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1098890203824128e+16, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3200.9990234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10983.6708984375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.416926812353331e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1046108392065806.2, + "step": 329, + "teaching_rms": 58366182933843.71, + "traffic_rms": 1046108392065806.2 + }, + { + "batch_loss": 20720972.0, + "innovation_rms": 61748912538948.24, + "instruction_rms": 9.826313414272176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.727261503287774e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.248299541727453e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9982584180178944.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3773.568115234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7989.7958984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.516752589730611e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.486727590055772e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5539033308266496e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276431022456832.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29679077556224.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9982584180178944.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3773.467529296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7989.7958984375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.516752589730611e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1106742545406956.9, + "step": 330, + "teaching_rms": 61748912538948.24, + "traffic_rms": 1106742545406956.9 + }, + { + "batch_loss": 19183538.0, + "innovation_rms": 65327700880713.73, + "instruction_rms": 8.641511412317985, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0280397127958628e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.1451163550321692e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8977808793534464.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4182.3544921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6138.85986328125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.606530720615629e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.86271727348482e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.759898208680018e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276567924539392.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32527951593472.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8977808793534464.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4182.25439453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6138.85986328125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.606530720615629e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1170891365019200.2, + "step": 331, + "teaching_rms": 65327700880713.73, + "traffic_rms": 1170891365019200.2 + }, + { + "batch_loss": 16782382.0, + "innovation_rms": 69113933636991.92, + "instruction_rms": 7.261149280097987, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0715930396297443e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.434832700366166e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8073420751241216.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4455.75439453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5764.16357421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.687264938865459e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.260500651765924e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9778327520188826e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276691136413696.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38147001942016.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8073420751241216.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4455.654296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5764.16357421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.687264938865459e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1238758370179005.0, + "step": 332, + "teaching_rms": 69113933636991.92, + "traffic_rms": 1238758370179005.0 + }, + { + "batch_loss": 21004004.0, + "innovation_rms": 73119611106027.31, + "instruction_rms": 5.713199890452633, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1133802517222549e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.8479846659873277e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7259391304663040.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4605.2353515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5932.01123046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.759858905599181e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.68134078888149e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.208399309571359e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276802000257024.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44386196914176.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7259391304663040.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4605.13525390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5932.01123046875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.759858905599181e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1310559106440884.8, + "step": 333, + "teaching_rms": 73119611106027.31, + "traffic_rms": 1310559106440884.8 + }, + { + "batch_loss": 24629904.0, + "innovation_rms": 77357471405707.62, + "instruction_rms": 6.403216572412973, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1535361052184947e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 2.4170712727769938e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6526692432543744.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4654.66845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5492.53857421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.825125658125926e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.126573591459267e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.452330430959452e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276901589811200.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51547677392896.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6526692432543744.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4654.568359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5492.53857421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.825125658125926e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1386521722649995.2, + "step": 334, + "teaching_rms": 77357471405707.62, + "traffic_rms": 1386521722649995.2 + }, + { + "batch_loss": 27059714.0, + "innovation_rms": 81840962133856.19, + "instruction_rms": 6.809218338658145, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1968632669085893e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 3.1950121362770406e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5867198056759296.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4585.8486328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5233.1416015625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.883797488369664e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.597613993525248e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.7104005846859776e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4276991247253504.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59111358070784.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5867198056759296.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4585.7490234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5233.1416015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.883797488369664e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1466887184190599.0, + "step": 335, + "teaching_rms": 81840962133856.19, + "traffic_rms": 1466887184190599.0 + }, + { + "batch_loss": 2730127.0, + "innovation_rms": 86584340764456.3, + "instruction_rms": 2.3964464821220197, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1702666963591994e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.679514704839454e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5273594384875520.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4693.880859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4709.60205078125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.9365332443136e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.095956643698442e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.9834295941083955e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277072046325760.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85797348835328.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5273594384875520.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4693.9755859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4709.60205078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.9365332443136e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1551910995883213.0, + "step": 336, + "teaching_rms": 86584340764456.3, + "traffic_rms": 1551910995883213.0 + }, + { + "batch_loss": 1997740.375, + "innovation_rms": 91602626536366.33, + "instruction_rms": 2.1842504682656427, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1440361136482775e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.724884820423001e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4739298305769472.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4838.9306640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4238.369140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.983926060941312e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.623184838722847e+17, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.272284011828347e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277144792334336.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77217614790656.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4739298305769472.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4839.025390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4238.369140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.983926060941312e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1641862811492583.5, + "step": 337, + "teaching_rms": 91602626536366.33, + "traffic_rms": 1641862811492583.5 + }, + { + "batch_loss": 1219564.625, + "innovation_rms": 96911790535450.97, + "instruction_rms": 2.1950427761294207, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1185131590288063e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.752704814812104e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4258384375185408.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4969.42724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3814.217529296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.026509802687693e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0180973272246518e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.5778808992628736e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277210290585600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69495850795008.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4258384375185408.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4969.52197265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3814.217529296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.026509802687693e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1737028558903593.2, + "step": 338, + "teaching_rms": 96911790535450.97, + "traffic_rms": 1737028558903593.2 + }, + { + "batch_loss": 752416.125, + "innovation_rms": 102528694936077.31, + "instruction_rms": 2.1204606656289937, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0939954219199129e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.706933142428314e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3825519317483520.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5086.82421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432.442626953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.064765076393165e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0771092157990175e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.901191606215639e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277269077950464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62546262360064.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3825519317483520.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5086.9189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432.442626953125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.064765076393165e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1837710483889610.5, + "step": 339, + "teaching_rms": 102528694936077.31, + "traffic_rms": 1837710483889610.5 + }, + { + "batch_loss": 158436.4375, + "innovation_rms": 108471149644042.69, + "instruction_rms": 1.7055180653322093, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0706888012670614e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.568637831508334e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3435902701404160.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5192.4306640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3088.81103515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.099123955767706e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1395416163279176e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.243242114474312e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277321959735296.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56291636543488.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3435902701404160.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5192.525390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3088.81103515625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.099123955767706e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1944228063263437.8, + "step": 340, + "teaching_rms": 108471149644042.69, + "traffic_rms": 1944228063263437.8 + }, + { + "batch_loss": 108873.734375, + "innovation_rms": 114758130312481.31, + "instruction_rms": 1.838158443528209, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0487778742862296e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.429021019596418e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3085213252976640.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5287.4248046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2779.51171875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.129975994345062e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2055927845017354e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.605118878966088e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277369741246464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 154186725457920.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3085213252976640.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5287.51953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2779.51171875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.129975994345062e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2056919717605552.2, + "step": 341, + "teaching_rms": 114758130312481.31, + "traffic_rms": 2056919717605552.2 + }, + { + "batch_loss": 64755.39453125, + "innovation_rms": 121409506059358.61, + "instruction_rms": 1.6461324731154103, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0284224483954451e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.27805500551809e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2769561879314432.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5372.86669921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2501.114501953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.157671661456589e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.275472520882946e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.987970889341993e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277412690919424.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 432938005037056.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2769561879314432.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5372.96142578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2501.114501953125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.157671661456589e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2176143302267773.8, + "step": 342, + "teaching_rms": 121409506059358.61, + "traffic_rms": 2176143302267773.8 + }, + { + "batch_loss": 58709.11328125, + "innovation_rms": 128446351910422.81, + "instruction_rms": 1.933254502910819, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0097099919437207e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.0945764305892434e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2485448182071296.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5449.71044921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2250.531982421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.182526207701811e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.34940272066193e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.393013793145487e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277451345625088.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 729083449180160.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2485448182071296.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5449.80517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2250.531982421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.182526207701811e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2302277522063243.5, + "step": 343, + "teaching_rms": 128446351910422.81, + "traffic_rms": 2302277522063243.5 + }, + { + "batch_loss": 28854.4296875, + "innovation_rms": 135891074843801.89, + "instruction_rms": 1.8946737379644554, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.927257918489997e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 4.9463720231205315e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2229720728993792.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5518.8154296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2024.9853515625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204823530418995e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4276181295712174e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.821535393370604e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277486242234368.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1032161708736512.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2229720728993792.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5518.91015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2024.9853515625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.204823530418995e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2435722806728885.5, + "step": 344, + "teaching_rms": 135891074843801.89, + "traffic_rms": 2435722806728885.5 + }, + { + "batch_loss": 8650.5654296875, + "innovation_rms": 143767294364729.84, + "instruction_rms": 1.8029825268904796, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.774932105343491e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 7.124954611556712e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1999543767924736.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5580.95458984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1821.97314453125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.224818750665523e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5103670997997322e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.27489496126718e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277517649182720.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1348565238546432.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1999543767924736.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5581.04931640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1821.97314453125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.224818750665523e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2576902901773972.0, + "step": 345, + "teaching_rms": 143767294364729.84, + "traffic_rms": 2576902901773972.0 + }, + { + "batch_loss": 1053.257568359375, + "innovation_rms": 152100065492328.66, + "instruction_rms": 1.7903294045509361, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.640270254360463e+20, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 9.882285774046675e+32, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1792364477677568.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5636.82421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1639.2437744140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.24274250818519e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.597912414626513e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.754532169872835e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277545834905600.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1704760298176512.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1792364477677568.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5636.9189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1639.2437744140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.24274250818519e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2726266270873539.5, + "step": 346, + "teaching_rms": 152100065492328.66, + "traffic_rms": 2726266270873539.5 + }, + { + "batch_loss": 817.9069213867188, + "innovation_rms": 160915838956355.16, + "instruction_rms": 1.8318252292060744, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.003726959838759e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.2276024662448228e+33, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1605885184966656.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5687.05029296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1474.77099609375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.258801390904934e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6905321130544333e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.261971217181573e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277571067838464.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2138543002484736.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1605885184966656.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5687.14501953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1474.77099609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.258801390904934e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2884287041155273.0, + "step": 347, + "teaching_rms": 160915838956355.16, + "traffic_rms": 2884287041155273.0 + }, + { + "batch_loss": 924.5974731445312, + "innovation_rms": 170243818104436.75, + "instruction_rms": 1.2088213845551425, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0617996807650505e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.882622893280461e+16, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 9, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2446919372636160.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5763.31396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1326.730712890625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.273181800405402e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7885203144439235e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.798822887728087e+17, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277593884852224.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2734754390605824.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2446919372636160.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5763.294921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1326.730712890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.273181800405402e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3051467404314031.5, + "step": 348, + "teaching_rms": 170243818104436.75, + "traffic_rms": 3051467404314031.5 + }, + { + "batch_loss": 1267.814697265625, + "innovation_rms": 180110399080661.25, + "instruction_rms": 1.1731698287971755, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.123232298119593e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.8825975529734144e+16, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 9, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6300574752440320.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5882.662109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193.4813232421875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.286051240411136e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.892188180585644e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0366790737340662e+18, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277614285946880.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4081309852893184.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6300574752440320.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5882.64306640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193.4813232421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.286051240411136e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3228337267183205.5, + "step": 349, + "teaching_rms": 180110399080661.25, + "traffic_rms": 3228337267183205.5 + }, + { + "batch_loss": 670.827880859375, + "innovation_rms": 190548693622654.94, + "instruction_rms": 1.2536544002407344, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.188219170373805e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.8697704183955456e+16, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 9, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5670413929545728.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5990.0166015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1073.544921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.297560893770957e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.001865015212114e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.096768071386792e+18, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277632807993344.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3673178706542592.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5670413929545728.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5989.99755859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1073.544921875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.297560893770957e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3415459253099610.0, + "step": 350, + "teaching_rms": 190548693622654.94, + "traffic_rms": 3415459253099610.0 + }, + { + "batch_loss": 715.483642578125, + "innovation_rms": 201595145321215.34, + "instruction_rms": 1.771657033484916, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2569660557356622e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.857259822907392e+16, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 9, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5103212495372288.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6086.57568359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 965.5914306640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.307846481451418e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1178989511924777e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1603399095594516e+18, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277649450991616.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3305860755357696.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5103212495372288.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6086.556640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 965.5914306640625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.307846481451418e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3613427734797328.5, + "step": 351, + "teaching_rms": 201595145321215.34, + "traffic_rms": 3613427734797328.5 + }, + { + "batch_loss": 618.78369140625, + "innovation_rms": 213280226230848.53, + "instruction_rms": 1.2926257785505966, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3296904639934177e+21, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": false, + "max_abs_over_finite_tensors": 1.8274188196315136e+16, + "nonfinite_indices": [ + 3, + 4, + 7, + 8, + 9, + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18 + ] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4592679933444096.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6173.41796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868.4236450195312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.317030839517184e+16, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2406585997999473e+18, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2275966235136492e+18, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4277664483377152.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4048775475625984.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4592679933444096.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6173.39892578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868.4236450195312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.317030839517184e+16, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251719295978546, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3822870723628354.5, + "step": 352, + "teaching_rms": 213280226230848.53, + "traffic_rms": 3822870723628354.5 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} |