diff options
Diffstat (limited to 'results/kp_innovation_stability_grid/margin_0p01.json')
| -rw-r--r-- | results/kp_innovation_stability_grid/margin_0p01.json | 32986 |
1 files changed, 32986 insertions, 0 deletions
diff --git a/results/kp_innovation_stability_grid/margin_0p01.json b/results/kp_innovation_stability_grid/margin_0p01.json new file mode 100644 index 0000000..506dedb --- /dev/null +++ b/results/kp_innovation_stability_grid/margin_0p01.json @@ -0,0 +1,32986 @@ +{ + "first_any_nonfinite_groups": {}, + "first_any_nonfinite_step": null, + "first_nonfinite_by_group": {}, + "first_nonfinite_by_metric": {}, + "first_nonfinite_metrics": [], + "first_training_failure_groups": {}, + "first_training_failure_step": null, + "initial_parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4011031687259674, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3956821858882904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1162850856781006, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 0.011125171557068825, + "max_applied_stability_margin": 0.0111251724883914, + "max_positive_residual_soma_slope": 0.0, + "min_residual_soma_slope": -0.011125171557068825, + "mse": 8.080039934493075e-05, + "observations": 64, + "residual_traffic_rms_ratio": 0.5331967502228365, + "stability_margin": 0.01 + }, + "first_mse": 8.080039934493075e-05, + "last_mse": 8.080039934493075e-05, + "post_warmup_traffic_residual_rms_ratio": 0.5331967502228365, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_mixed_traffic_nonfinite_diagnosis_v1", + "provenance": { + "git_commit": "0ae690e7b3144da2c5b3384cd7169e86529fd26f", + "git_tracked_dirty": false + }, + "rule": "innovation", + "scope": "training_only_no_validation_or_test_evaluation", + "split": { + "dataset": "cifar10", + "input_layout": "NCHW", + "input_shape": [ + 3, + 32, + 32 + ], + "loader_seed": 0, + "normalization_mean": [ + 0.49140000343322754, + 0.4821999967098236, + 0.4465000033378601 + ], + "normalization_std": [ + 0.24699999392032623, + 0.2434999942779541, + 0.26159998774528503 + ], + "split_from_training_only": true, + "split_seed": 2027, + "test_examples": 10000, + "train_examples": 45000, + "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5", + "validation_class_counts": { + "0": 500, + "1": 500, + "2": 500, + "3": 500, + "4": 500, + "5": 500, + "6": 500, + "7": 500, + "8": 500, + "9": 500 + }, + "validation_examples": 5000, + "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b" + }, + "stability_margin": 0.01, + "test_evaluations": 0, + "traffic_calibration": { + "instruction_rms": [ + 0.008471464738249779, + 0.006617916747927666, + 0.003988116513937712, + 0.0043865954503417015, + 0.002713605063036084, + 0.003158062230795622, + 0.002004207344725728, + 0.003757704282179475, + 0.0027085731271654367, + 0.0030200635083019733, + 0.002056022873148322, + 0.002389939734712243, + 0.0016842411132529378, + 0.0034010226372629404, + 0.002549725119024515, + 0.0030542907770723104, + 0.0020918569061905146, + 0.002605808200314641, + 0.0018916188273578882 + ], + "realized_traffic_instruction_rms_ratio": [ + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 4.0, + 3.999999523162842, + 4.0, + 4.000000476837158, + 3.999999761581421, + 4.0, + 4.0, + 3.999999761581421, + 4.0, + 4.000000476837158, + 4.000000476837158, + 4.0, + 4.0, + 4.0 + ], + "target_ratio": 4.0, + "traffic_gain": [ + 0.0443243607878685, + 0.035817068070173264, + 0.013778294436633587, + 0.022762222215533257, + 0.007214079145342112, + 0.017218582332134247, + 0.004616298712790012, + 0.02042512409389019, + 0.007327491417527199, + 0.01611751690506935, + 0.004703814629465342, + 0.01256294921040535, + 0.003324519144371152, + 0.01805141195654869, + 0.006063441745936871, + 0.016296451911330223, + 0.0044285994954407215, + 0.013874167576432228, + 0.0036007820162922144 + ], + "unscaled_traffic_rms": [ + 0.7644973993301392, + 0.7390796542167664, + 1.157796859741211, + 0.7708553671836853, + 1.504616141319275, + 0.7336404919624329, + 1.736635684967041, + 0.7358984351158142, + 1.478581428527832, + 0.7495108246803284, + 1.7483876943588257, + 0.7609485983848572, + 2.0264477729797363, + 0.7536302804946899, + 1.6820316314697266, + 0.7496824264526367, + 1.8894071578979492, + 0.7512690424919128, + 2.10134220123291 + ] + }, + "trajectory": [ + { + "batch_loss": 2.9134020805358887, + "innovation_rms": 0.010081594228070463, + "instruction_rms": 0.00421539791368719, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40285658836364746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9875131845474243, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.376714706420898, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4058851897716522, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6322960257530212, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9460533857345581, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39024874567985535, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9024875164031982, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0241279602050781, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.568367958068848, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.37673568725586, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3794170618057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.632293164730072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.203450083732605, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01731826256776222, + "step": 1, + "teaching_rms": 0.010081594228070463, + "traffic_rms": 0.016800932580503868 + }, + { + "batch_loss": 2.3531951904296875, + "innovation_rms": 0.0959486766972197, + "instruction_rms": 0.0957219530896388, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6021227836608887, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.693324327468872, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.437236785888672, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40752050280570984, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5040203332901001, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7448183298110962, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7434362173080444, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.591897010803223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.011120319366455, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.987762451171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.437277793884277, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38057780265808105, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5040148496627808, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8815804719924927, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.09608516507643175, + "step": 2, + "teaching_rms": 0.0959486766972197, + "traffic_rms": 0.008262247913726184 + }, + { + "batch_loss": 2.476445436477661, + "innovation_rms": 16.099985356110803, + "instruction_rms": 16.099985018739243, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2091940641403198, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.727959394454956, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.644065856933594, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41007092595100403, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36438536643981934, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.224384784698486, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.47871971130371, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.10513305664062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.049942016601562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.5108871459961, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.59001159667969, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40170571208000183, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36437755823135376, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.532761573791504, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16.099986497611177, + "step": 3, + "teaching_rms": 16.099985356110803, + "traffic_rms": 0.004914635765512741 + }, + { + "batch_loss": 4.717585563659668, + "innovation_rms": 0.03450619975424598, + "instruction_rms": 0.008064917344835476, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.71405792236328, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7394.06103515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.278602600097656, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9904522895812988, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.045193672180176, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.15224552154541, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.664196014404297, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.42710876464844, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.688008308410645, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.08735656738281, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.22111511230469, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7778841257095337, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.045267105102539, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.15487289428711, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.09613428778531108, + "step": 4, + "teaching_rms": 0.03450619975424598, + "traffic_rms": 0.09580256463310743 + }, + { + "batch_loss": 3.9219820499420166, + "innovation_rms": 0.03796869601488627, + "instruction_rms": 0.030948467456718382, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.6272201538086, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24639.080078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.349918365478516, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.891523003578186, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.010706901550293, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.687237739562988, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.99488067626953, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.7504425048828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.351335525512695, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.95751190185547, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.60025405883789, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.678963541984558, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.010793685913086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.114898681640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.08668192033364577, + "step": 5, + "teaching_rms": 0.03796869601488627, + "traffic_rms": 0.08097444324816802 + }, + { + "batch_loss": 3.992255926132202, + "innovation_rms": 0.13589132503655507, + "instruction_rms": 0.13523142019951254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.2039794921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22438.4375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.05703735351562, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.702415943145752, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.108929634094238, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.86840534210205, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.58065414428711, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 173.16921997070312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.080402374267578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 113.04670715332031, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.05722045898438, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4898664951324463, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.109028816223145, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.478689193725586, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.13802826740103066, + "step": 6, + "teaching_rms": 0.13589132503655507, + "traffic_rms": 0.027650310279709857 + }, + { + "batch_loss": 3.608405590057373, + "innovation_rms": 0.013078816724299534, + "instruction_rms": 0.006476119296010742, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 105.34825134277344, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 31767.5859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.2530517578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.432016134262085, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.2960028648376465, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.731307983398438, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.48666000366211, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 168.5965576171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.189857482910156, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.73487854003906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.25325393676758, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.219477891921997, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.29611349105835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.30579948425293, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02595734653270197, + "step": 7, + "teaching_rms": 0.013078816724299534, + "traffic_rms": 0.025146052477695365 + }, + { + "batch_loss": 4.016349792480469, + "innovation_rms": 0.02493239233244421, + "instruction_rms": 0.008115198625113972, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.33341979980469, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 49775.22265625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.83855438232422, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.079925060272217, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.479089736938477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.307743072509766, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.20425415039062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.7844696044922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.253345489501953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.9995574951172, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.83851623535156, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.867398977279663, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.47921085357666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.6498908996582, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.030206046833929266, + "step": 8, + "teaching_rms": 0.02493239233244421, + "traffic_rms": 0.02910467753527998 + }, + { + "batch_loss": 4.285613536834717, + "innovation_rms": 0.02372878930757595, + "instruction_rms": 0.006981654984723471, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.53214263916016, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75540.9140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.85408401489258, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.656559944152832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.7663493156433105, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.131980895996094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.75064086914062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.11331176757812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.704914093017578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 166.79315185546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.854042053222656, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.444046974182129, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.766479969024658, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.559104919433594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019655422344281314, + "step": 9, + "teaching_rms": 0.02372878930757595, + "traffic_rms": 0.01838870649647694 + }, + { + "batch_loss": 3.5335745811462402, + "innovation_rms": 0.012893581678316203, + "instruction_rms": 0.0065881682828800876, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 91.0907211303711, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 103767.53125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.45674133300781, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.17546272277832, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.189027309417725, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.596832275390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.24238586425781, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.81564331054688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.96562385559082, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.11383056640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.45669937133789, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.962963581085205, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.189166069030762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.07658767700195, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018757318966569125, + "step": 10, + "teaching_rms": 0.012893581678316203, + "traffic_rms": 0.017571618449407387 + }, + { + "batch_loss": 3.4827187061309814, + "innovation_rms": 0.020468156119700223, + "instruction_rms": 0.011324252957045394, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.9716567993164, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 202009.828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.61726760864258, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.642416477203369, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.669539451599121, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.61492156982422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.20376586914062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.96839141845703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.715017318725586, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54.50912857055664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.61722183227539, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.429932117462158, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.6696858406066895, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 45.2414436340332, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.022907930117821326, + "step": 11, + "teaching_rms": 0.020468156119700223, + "traffic_rms": 0.019916931885873533 + }, + { + "batch_loss": 4.213178634643555, + "innovation_rms": 0.03997673600102462, + "instruction_rms": 0.03800572644876113, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 87.5232162475586, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 265945.28125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.254390716552734, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.060915946960449, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.184996604919434, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.230892181396484, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 107.7209243774414, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.17154693603516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.990610122680664, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 110.44788360595703, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.25434112548828, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.848447322845459, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.185149669647217, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.088958740234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03976469849141744, + "step": 12, + "teaching_rms": 0.03997673600102462, + "traffic_rms": 0.011710525315856358 + }, + { + "batch_loss": 3.547884225845337, + "innovation_rms": 0.04959049681590618, + "instruction_rms": 0.0414100338622459, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.97247314453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 240780.734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.326480865478516, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.43750524520874, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.7658908367156982, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.4849967956543, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.08636474609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103.65438842773438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.97296905517578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 225.1593017578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.32643508911133, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.225052356719971, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.766049861907959, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.6511344909668, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.045987377786838216, + "step": 13, + "teaching_rms": 0.04959049681590618, + "traffic_rms": 0.02001155970961695 + }, + { + "batch_loss": 4.099434852600098, + "innovation_rms": 0.030805955385111612, + "instruction_rms": 0.02345406785659076, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.50049591064453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1557456.625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.88054275512695, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.773758888244629, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3625361919403076, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 41.42168045043945, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.4152603149414, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.28894805908203, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.73145294189453, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.84249877929688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.88072204589844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.561322212219238, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3627004623413086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.956565856933594, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.03904105132729888, + "step": 14, + "teaching_rms": 0.030805955385111612, + "traffic_rms": 0.03121298692662914 + }, + { + "batch_loss": 4.49403715133667, + "innovation_rms": 0.06735415991048833, + "instruction_rms": 0.06529582666987746, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.79850769042969, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1402020.5, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.99201965332031, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.073694229125977, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9993536472320557, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.069618225097656, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.81491088867188, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.99649810791016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.31565856933594, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 116.09706115722656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.99220275878906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.861274719238281, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9995226860046387, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.03092575073242, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.07368662364952919, + "step": 15, + "teaching_rms": 0.06735415991048833, + "traffic_rms": 0.03414934946054371 + }, + { + "batch_loss": 4.082787036895752, + "innovation_rms": 0.032925186108314033, + "instruction_rms": 0.024781017940796676, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 132.39004516601562, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1773419.125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.65672302246094, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.3423752784729, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.686810255050659, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 46.45232391357422, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 143.41958618164062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.04668426513672, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.7414436340332, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 140.2861328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 213.65672302246094, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.1299729347229, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.686983585357666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.89729690551758, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.030638946332581993, + "step": 16, + "teaching_rms": 0.032925186108314033, + "traffic_rms": 0.01802154534716437 + }, + { + "batch_loss": 4.7557692527771, + "innovation_rms": 145.33944056094387, + "instruction_rms": 145.33943888737656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.04461669921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2519415.75, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 434.11712646484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.5841145515441895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4173948764801025, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.59629440307617, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.30726623535156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.87674713134766, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.675392150878906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.01974487304688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 434.1169128417969, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.371730327606201, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.417572021484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.57646179199219, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 145.33944266080783, + "step": 17, + "teaching_rms": 145.33944056094387, + "traffic_rms": 0.033126767453312854 + }, + { + "batch_loss": 4.059061050415039, + "innovation_rms": 4.436981768444155, + "instruction_rms": 4.436955053413805, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 146.59840393066406, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2320961.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 390.700927734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.801482677459717, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1736795902252197, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.24465942382812, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 156.5061798095703, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.989070892333984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.11990737915039, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.81776428222656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 390.7007141113281, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.58911657333374, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1738603115081787, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48573303222656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4.437048537452248, + "step": 18, + "teaching_rms": 4.436981768444155, + "traffic_rms": 0.02880254522379143 + }, + { + "batch_loss": 7.028702735900879, + "innovation_rms": 0.5707960384057782, + "instruction_rms": 0.5706020323831402, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 157.51519775390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2097844.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.6224365234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.990458011627197, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.373170375823975, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.40690612792969, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.08518981933594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.7901611328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.819969177246094, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.4359893798828, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.6222229003906, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.778110027313232, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.373010158538818, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 118.64795684814453, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.5712966650351449, + "step": 19, + "teaching_rms": 0.5707960384057782, + "traffic_rms": 0.028175225166828882 + }, + { + "batch_loss": 7.875793933868408, + "innovation_rms": 0.2271058321131441, + "instruction_rms": 0.22207768968274055, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.55233764648438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9715103.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 344.1450500488281, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.149005889892578, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.581125020980835, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.0508575439453, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.10630798339844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50.211143493652344, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.920204162597656, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 365.2312927246094, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 344.14483642578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.936676979064941, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.5809621810913086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.29188537597656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.22514969636042226, + "step": 20, + "teaching_rms": 0.2271058321131441, + "traffic_rms": 0.03707684169613999 + }, + { + "batch_loss": 8.353155136108398, + "innovation_rms": 0.0783114862106525, + "instruction_rms": 0.06808342204504879, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.81932067871094, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778082.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309.7271423339844, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.291596412658691, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.879405975341797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 178.52891540527344, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.62530517578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 264.5400390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.43459701538086, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 328.7081604003906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 309.7269287109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.079286575317383, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.879554748535156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 178.76991271972656, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.08084429382613317, + "step": 21, + "teaching_rms": 0.0783114862106525, + "traffic_rms": 0.043603594773344745 + }, + { + "batch_loss": 8.006993293762207, + "innovation_rms": 0.06563890372502573, + "instruction_rms": 0.05585668007859487, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.51760864257812, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15499406.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 294.6337585449219, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.4197359085083, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.830373764038086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 204.15737915039062, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 175.69241333007812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.08602905273438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.70216369628906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 131.5056915283203, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 294.6338806152344, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.20744514465332, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.830523490905762, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 204.3983612060547, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06191756154683682, + "step": 22, + "teaching_rms": 0.06563890372502573, + "traffic_rms": 0.02674067065547566 + }, + { + "batch_loss": 7.177463531494141, + "innovation_rms": 0.040374080690925795, + "instruction_rms": 0.005132023086861793, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 209.08587646484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 37142528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 265.1800537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.534996032714844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.907163143157959, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 227.220947265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 179.35279846191406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.2774200439453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.15597534179688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257.1410827636719, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 265.1801452636719, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.322725296020508, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.907314777374268, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 227.4619140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04318859118134587, + "step": 23, + "teaching_rms": 0.040374080690925795, + "traffic_rms": 0.04288969857696361 + }, + { + "batch_loss": 8.723922729492188, + "innovation_rms": 2.110580508124699, + "instruction_rms": 2.1105327595176826, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.71315002441406, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 33462276.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.6568145751953, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.63864517211914, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.104203224182129, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.97589111328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.64715576171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.84967041015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.86441040039062, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.38978576660156, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 238.65689086914062, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.42639446258545, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.10435676574707, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 248.2168426513672, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.1106559901050965, + "step": 24, + "teaching_rms": 2.110580508124699, + "traffic_rms": 0.022839290281816225 + }, + { + "batch_loss": 8.477272033691406, + "innovation_rms": 0.04854714509472903, + "instruction_rms": 0.017244466314942093, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.68807983398438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64694784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.78311157226562, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.731842994689941, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.389216899871826, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.6532897949219, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 185.62393188476562, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 173.564697265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.9020004272461, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.2144775390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.783203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.519612312316895, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.389371871948242, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.89422607421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.04314164787736114, + "step": 25, + "teaching_rms": 0.04854714509472903, + "traffic_rms": 0.03955776604945336 + }, + { + "batch_loss": 7.216635227203369, + "innovation_rms": 0.07007175965124711, + "instruction_rms": 0.04919473790585064, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 214.25401306152344, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 58233340.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 711.973388671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.818131446838379, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.749886989593506, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 283.4603271484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.3030242919922, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 156.2311248779297, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 96.32251739501953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384.0735168457031, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 711.9732666015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.605920791625977, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.750042915344238, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 283.70123291015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.06113111947517135, + "step": 26, + "teaching_rms": 0.07007175965124711, + "traffic_rms": 0.03630352563671873 + }, + { + "batch_loss": 7.899499416351318, + "innovation_rms": 0.38761681296120054, + "instruction_rms": 0.3851827597454544, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.9579620361328, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 96634368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858.9400634765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.895745277404785, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.138795375823975, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.5838317871094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.71420288085938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 141.8861541748047, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.20098114013672, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 490.28436279296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858.93994140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.683554649353027, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.138952255249023, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.82470703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3870710718147605, + "step": 27, + "teaching_rms": 0.38761681296120054, + "traffic_rms": 0.038203675397082124 + }, + { + "batch_loss": 4.937027931213379, + "innovation_rms": 2.7671832026716734, + "instruction_rms": 2.7659740098264276, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 220.73883056640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 142178368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 773.0374145507812, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.965508460998535, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.624398231506348, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 312.19287109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.88426208496094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.75308990478516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 105.59813690185547, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 558.0458984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 773.0372924804688, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.753337860107422, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.624556064605713, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 312.4337158203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.7668166986973253, + "step": 28, + "teaching_rms": 2.7671832026716734, + "traffic_rms": 0.06827873057653018 + }, + { + "batch_loss": 4.402019023895264, + "innovation_rms": 0.2195155082531269, + "instruction_rms": 0.21369994660808156, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 228.6546173095703, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 127972888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 695.7169799804688, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.028210639953613, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1613945960998535, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 324.5231018066406, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.84068298339844, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.92386627197266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 120.13654327392578, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 639.00146484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 695.7168579101562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.816060066223145, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.161553382873535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 324.763916015625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.21610634412383697, + "step": 29, + "teaching_rms": 0.2195155082531269, + "traffic_rms": 0.032169402711184604 + }, + { + "batch_loss": 6.787515640258789, + "innovation_rms": 0.33281327294321084, + "instruction_rms": 0.32876799880718566, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1129.2327880859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 288686272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 626.1217041015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.085094451904297, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.737262725830078, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 335.6170654296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.60145568847656, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130.76242065429688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 143.3428497314453, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.48004150390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 626.12158203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.872963905334473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.737422466278076, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 335.85784912109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.332134013080419, + "step": 30, + "teaching_rms": 0.33281327294321084, + "traffic_rms": 0.047183348495037525 + }, + { + "batch_loss": 5.5442214012146, + "innovation_rms": 0.4912555781977062, + "instruction_rms": 0.49089101483332626, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1106.76171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 272572288.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1226.7703857421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.136213302612305, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.36289381980896, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.7355041503906, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.75120544433594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 121.81983184814453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 164.22853088378906, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 421.63201904296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1226.7705078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.924102783203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3630542755126953, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 351.87677001953125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.49114790565317035, + "step": 31, + "teaching_rms": 0.4912555781977062, + "traffic_rms": 0.015900015400568776 + }, + { + "batch_loss": 3.944164514541626, + "innovation_rms": 0.7408998999258656, + "instruction_rms": 0.7295190653287797, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1015.7631225585938, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 371731520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1104.0810546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.18210506439209, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.025928258895874, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 402.4450988769531, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 208.2048797607422, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147.40371704101562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 184.5114288330078, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 897.6857299804688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1104.0811767578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.970015525817871, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0260891914367676, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 402.5863342285156, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.7353243364072495, + "step": 32, + "teaching_rms": 0.7408998999258656, + "traffic_rms": 0.09221894603806628 + }, + { + "batch_loss": 27.334157943725586, + "innovation_rms": 6.919107846305785, + "instruction_rms": 6.915842468263619, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1824.998046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 781770112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 993.6495971679688, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.23337459564209, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.058635234832764, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 448.0796813964844, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 216.38540649414062, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.01849365234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 134.91815185546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1696.5177001953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 993.6497802734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.021306037902832, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.058505535125732, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 448.2209167480469, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6.917091993786396, + "step": 33, + "teaching_rms": 6.919107846305785, + "traffic_rms": 0.13153944130677583 + }, + { + "batch_loss": 57.09222412109375, + "innovation_rms": 320.5067592275056, + "instruction_rms": 320.50675463091756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1675.472412109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 703593280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 894.2514038085938, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.27538013458252, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.599193572998047, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 489.1463317871094, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 223.7478790283203, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 180.01663208007812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 142.51890563964844, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1526.8658447265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 894.2515869140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.063332557678223, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.5994815826416, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 489.2875671386719, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 320.5067579891325, + "step": 34, + "teaching_rms": 320.5067592275056, + "traffic_rms": 0.04655533606155763 + }, + { + "batch_loss": 28.170204162597656, + "innovation_rms": 411.15256489118724, + "instruction_rms": 411.152551480505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1533.6815185546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 633281280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 804.7840576171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.302918434143066, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.13445281982422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 526.1014404296875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 232.9495849609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 229.02017211914062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 275.2448425292969, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1374.17919921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 804.7842407226562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.09089183807373, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24.134742736816406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 526.24267578125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 411.152559150942, + "step": 35, + "teaching_rms": 411.15256489118724, + "traffic_rms": 0.07951965305321788 + }, + { + "batch_loss": 40.54689025878906, + "innovation_rms": 30700.072198084814, + "instruction_rms": 30700.07219778774, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1827.9171142578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 570065344.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 724.25537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.364463806152344, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.719635009765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 574.6162109375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 263.4642639160156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 418.968505859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 398.92095947265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1236.76123046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 724.2555541992188, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.152458190917969, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.719924926757812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 574.52880859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30700.07219791285, + "step": 36, + "teaching_rms": 30700.072198084814, + "traffic_rms": 0.08776080016254979 + }, + { + "batch_loss": 17.797855377197266, + "innovation_rms": 403.5931129836903, + "instruction_rms": 403.5931117737206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2045.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 848048768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 651.7723999023438, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.419178009033203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.214488983154297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 639.79345703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 290.927490234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 377.0716552734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 510.2294616699219, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1113.0850830078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 651.7725219726562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.207193374633789, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.214780807495117, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 639.7060546875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 403.59311228208264, + "step": 37, + "teaching_rms": 403.5931129836903, + "traffic_rms": 0.020257045830731447 + }, + { + "batch_loss": 27.424907684326172, + "innovation_rms": 729.2354463536564, + "instruction_rms": 729.2354460568572, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2213.6240234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 886117312.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 586.5311889648438, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.618478775024414, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.381807327270508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 698.4465942382812, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 315.6443786621094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 339.3644714355469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 610.4071044921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1001.7765502929688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 586.5313110351562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.3204345703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.382101058959961, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 698.3591918945312, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 729.2354464989188, + "step": 38, + "teaching_rms": 729.2354463536564, + "traffic_rms": 0.025419443265906824 + }, + { + "batch_loss": 12.860934257507324, + "innovation_rms": 1031.5539689372276, + "instruction_rms": 1031.553968168041, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2115.095703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 810429568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541.5380859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.987052917480469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.68574333190918, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 751.2274169921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 341.75408935546875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 331.179931640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 700.5670166015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 901.598876953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 541.5380249023438, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.689038276672363, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.686037063598633, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 751.1400146484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1031.5539687400435, + "step": 39, + "teaching_rms": 1031.5539689372276, + "traffic_rms": 0.034385276203166545 + }, + { + "batch_loss": 6.751663684844971, + "innovation_rms": 0.2720403663334642, + "instruction_rms": 0.09210476393446367, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1904.1337890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2727967488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487.3787841796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.218649864196777, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.315969467163086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 798.72265625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 365.9256591796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 298.0619201660156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 781.7109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2313.211669921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487.37872314453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.92066478729248, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.316264152526855, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 798.63525390625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27344411048185807, + "step": 40, + "teaching_rms": 0.2720403663334642, + "traffic_rms": 0.2574662555454934 + }, + { + "batch_loss": 18.007986068725586, + "innovation_rms": 539688.8008393119, + "instruction_rms": 539688.8008393111, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1721.8265380859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2455171072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438.63055419921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.28950023651123, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.708507537841797, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 841.4603881835938, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 387.6800842285156, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.5196533203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 854.7404174804688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2081.890380859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 438.6304626464844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.991544723510742, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.708802223205566, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 841.3729858398438, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 539688.8008393116, + "step": 41, + "teaching_rms": 539688.8008393119, + "traffic_rms": 0.021584998863311382 + }, + { + "batch_loss": 44.50376892089844, + "innovation_rms": 6237.225529188727, + "instruction_rms": 6237.22553079449, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889.501220703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2260607744.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2938.617431640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.218059539794922, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.285594940185547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 879.9158935546875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 407.2590637207031, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 334.0580749511719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 920.4669799804688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.7012939453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2938.61767578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.920133590698242, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.285890579223633, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 879.8285522460938, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 6237.225531137563, + "step": 42, + "teaching_rms": 6237.225529188727, + "traffic_rms": 0.0655516204610144 + }, + { + "batch_loss": 64.48465728759766, + "innovation_rms": 12469.311648506515, + "instruction_rms": 12469.311644763031, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4150.80322265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2239953920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2644.726318359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.043323516845703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.652318954467773, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 914.51708984375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 424.880126953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 783.7874145507812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 979.620849609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1794.2752685546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2644.7265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.745427131652832, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.652223587036133, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 914.4297485351562, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12469.311646306822, + "step": 43, + "teaching_rms": 12469.311648506515, + "traffic_rms": 0.19630033762948215 + }, + { + "batch_loss": 36.723934173583984, + "innovation_rms": 82598.21038097251, + "instruction_rms": 82598.21038090314, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4886.62646484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2078369536.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2380.19775390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.731266021728516, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.258593559265137, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 945.6489868164062, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 440.7391052246094, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 897.5376586914062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1032.859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3403.524169921875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2380.1982421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.433399200439453, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.258498191833496, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 945.5616455078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 82598.21038096452, + "step": 44, + "teaching_rms": 82598.21038097251, + "traffic_rms": 0.10123789419016044 + }, + { + "batch_loss": 55.940887451171875, + "innovation_rms": 1557.102263389444, + "instruction_rms": 1557.1022594297679, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4397.9638671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870532608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2141.693115234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.29916763305664, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.309737205505371, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010.1687622070312, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 455.0121765136719, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 807.7838745117188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1080.7740478515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3063.171630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2141.693603515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.00132942199707, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.30946159362793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1009.9336547851562, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1557.10226114387, + "step": 45, + "teaching_rms": 1557.102263389444, + "traffic_rms": 0.07314896991308704 + }, + { + "batch_loss": 35.345149993896484, + "innovation_rms": 598.1087116338985, + "instruction_rms": 598.1077044393702, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3948.345458984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1685703936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1926.5615234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.81010627746582, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.726997375488281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1202.824951171875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 468.327392578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 728.788818359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 727.7312622070312, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10629.5302734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1926.5618896484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.512296676635742, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.72672176361084, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1202.58984375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 598.1079733048625, + "step": 46, + "teaching_rms": 598.1087116338985, + "traffic_rms": 0.5671303034003861 + }, + { + "batch_loss": 133.76547241210938, + "innovation_rms": 158527.87474509006, + "instruction_rms": 158527.87474499148, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3553.510986328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1517133568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1733.994873046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.97441864013672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.88277244567871, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1376.2244873046875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 484.4061279296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 655.909912109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 974.443359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9566.5771484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1733.9952392578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.676637649536133, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.882568359375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1375.9893798828125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 158527.87474502518, + "step": 47, + "teaching_rms": 158527.87474509006, + "traffic_rms": 0.10355804713738204 + }, + { + "batch_loss": 6.397861957550049, + "innovation_rms": 10.514262981076765, + "instruction_rms": 10.481370877239348, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3422.409423828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11084286976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1897.5487060546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.122119903564453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.994735717773438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1532.23974609375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 514.4014892578125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 590.3189086914062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1196.4842529296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2220.40869140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1897.5484619140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.82436752319336, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.994531631469727, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1532.004638671875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10.492077872904646, + "step": 48, + "teaching_rms": 10.514262981076765, + "traffic_rms": 0.47388392427715326 + }, + { + "batch_loss": 70.169677734375, + "innovation_rms": 414.2098361758039, + "instruction_rms": 414.20911138057807, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3128.46533203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 21260943360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18922.873046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.16143798828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.295331954956055, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1955.88427734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 592.2528076171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 778.5133056640625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 799.4066772460938, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4612.81982421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18922.87109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.86371421813965, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.29512882232666, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1956.134521484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 414.20930694400096, + "step": 49, + "teaching_rms": 414.2098361758039, + "traffic_rms": 0.4025442846381394 + }, + { + "batch_loss": 131.86611938476562, + "innovation_rms": 1559.9342881903149, + "instruction_rms": 1559.9340093590085, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2815.038330078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24852635648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117730.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.19664192199707, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.264768600463867, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11768.453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 667.5974731445312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 753.4467163085938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 909.3094482421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6833.69580078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117730.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.898948669433594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.264877319335938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 11768.2900390625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1559.934104066611, + "step": 50, + "teaching_rms": 1559.9342881903149, + "traffic_rms": 0.5437015759217971 + }, + { + "batch_loss": 223.55084228515625, + "innovation_rms": 30.203763968395158, + "instruction_rms": 30.20322292355191, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2558.408935546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22367657984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105956.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.932937622070312, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.293128967285156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22364.103515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 735.4076538085938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 678.10205078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1034.61572265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6150.326171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105956.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.635272979736328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.293212890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 22363.939453125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 30.203563567185814, + "step": 51, + "teaching_rms": 30.203763968395158, + "traffic_rms": 0.14953460461273335 + }, + { + "batch_loss": 40.07557678222656, + "innovation_rms": 1086119.7085924225, + "instruction_rms": 1086119.7085924055, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5679.556640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20130891776.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95358.609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.695425033569336, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.06001853942871, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31899.96484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 796.4368286132812, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 610.2918090820312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1146.8643798828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5535.29345703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 95358.609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.397789001464844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.060100555419922, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 31899.80078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1086119.7085924202, + "step": 52, + "teaching_rms": 1086119.7085924225, + "traffic_rms": 0.177880450947875 + }, + { + "batch_loss": 59.78083801269531, + "innovation_rms": 415426.7442073996, + "instruction_rms": 415426.7442059848, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9762.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46754992128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85819.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.48148536682129, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.75715160369873, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40481.921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 961.7960205078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1653.592041015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1257.88720703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5431.63818359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85819.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.183879852294922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.757232666015625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 40481.7578125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 415426.74420637236, + "step": 53, + "teaching_rms": 415426.7442073996, + "traffic_rms": 0.5674636807020302 + }, + { + "batch_loss": 6.703315734863281, + "innovation_rms": 136680.54904510206, + "instruction_rms": 136680.54904436343, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10354.4453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64794472448.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77233.5546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.28876495361328, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.181466102600098, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48205.27734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1118.1910400390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1563.950439453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1357.782958984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10650.353515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77233.5546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.99118995666504, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.181547164916992, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 48205.11328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 136680.54904453852, + "step": 54, + "teaching_rms": 136680.54904510206, + "traffic_rms": 0.21878493385890838 + }, + { + "batch_loss": 7.703615188598633, + "innovation_rms": 2.9801227720386643, + "instruction_rms": 2.9603710825675247, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9321.42578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 58315042816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69505.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.115144729614258, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.76320743560791, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55155.81640625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1258.946533203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1407.5552978515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1450.0018310546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9585.318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69505.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.817598342895508, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.763289451599121, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 55155.65234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2.976886915431514, + "step": 55, + "teaching_rms": 2.9801227720386643, + "traffic_rms": 0.3131452517944176 + }, + { + "batch_loss": 2598.345947265625, + "innovation_rms": 2524010.6716852393, + "instruction_rms": 2524010.671685102, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8389.283203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 52483538944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62549.3203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.55067253112793, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273.21875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61410.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1385.62646484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1488.0880126953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 2312.680419921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8626.7861328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62549.3203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.516263961791992, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 273.2187805175781, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 61410.5859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2524010.671685152, + "step": 56, + "teaching_rms": 2524010.6716852393, + "traffic_rms": 0.5048717758003708 + }, + { + "batch_loss": 1448.7550048828125, + "innovation_rms": 881454.267846008, + "instruction_rms": 881454.2678458358, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7550.35498046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 47235186688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56288.24609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.58628463745117, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 240.35610961914062, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67039.578125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1499.638427734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2270.071533203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3089.091064453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7764.107421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56288.24609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.5518798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 240.35614013671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 67039.4140625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 881454.2678459082, + "step": 57, + "teaching_rms": 881454.267846008, + "traffic_rms": 0.3572339315994472 + }, + { + "batch_loss": 1959.366455078125, + "innovation_rms": 3200212.4507345036, + "instruction_rms": 3200212.4507344915, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7174.85595703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 42511667200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50652.71875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.75907135009766, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.72787475585938, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 72104.8515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1602.2491455078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2491.345458984375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 3787.860595703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6987.6962890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50652.71875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.72467041015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.7279052734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 72104.6875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3200212.4507344975, + "step": 58, + "teaching_rms": 3200212.4507345036, + "traffic_rms": 0.1913220749479154 + }, + { + "batch_loss": 856.2534790039062, + "innovation_rms": 8718.59403720605, + "instruction_rms": 8718.594035615735, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 20238.73828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38260498432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45580.234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.49893951416016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.39865112304688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 76662.875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1694.598876953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2371.378173828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4416.75341796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6288.92626953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45580.234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.46453857421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.39869689941406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 76662.7109375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8718.594036214286, + "step": 59, + "teaching_rms": 8718.59403720605, + "traffic_rms": 0.10663840972584693 + }, + { + "batch_loss": 1217.882568359375, + "innovation_rms": 9482.582014825799, + "instruction_rms": 9482.582014244454, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 64066.828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 104684822528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41014.54296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.00479888916016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.05860900878906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80764.328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1777.713623046875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2134.240234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 4982.7568359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5660.03369140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 41014.54296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 97.97040557861328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 155.0586700439453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 80764.1640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9482.582014698144, + "step": 60, + "teaching_rms": 9482.582014825799, + "traffic_rms": 0.09392252932514422 + }, + { + "batch_loss": 757.6437377929688, + "innovation_rms": 484.3608254424905, + "instruction_rms": 484.36054021563257, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 88372.7890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 386953641984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36905.0078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.92920684814453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.24411010742188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 84454.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.516845703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1920.816162109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5492.15966796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5094.0302734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36905.0078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.89482116699219, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 139.24415588378906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 84454.6640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 484.3606079873077, + "step": 61, + "teaching_rms": 484.3608254424905, + "traffic_rms": 0.2567074636976771 + }, + { + "batch_loss": 260.5519104003906, + "innovation_rms": 2024.5966125198725, + "instruction_rms": 2024.596597513427, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 117278.5390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 381824532480.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33206.0625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.45436096191406, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.25152587890625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 87775.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1989.920654296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1728.7344970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 5950.62255859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4584.626953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33206.0625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.41998291015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.25157928466797, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 87775.2734375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2024.596602206375, + "step": 62, + "teaching_rms": 2024.5966125198725, + "traffic_rms": 0.13825244350132954 + }, + { + "batch_loss": 85.90010070800781, + "innovation_rms": 1164.3514446415068, + "instruction_rms": 1164.3514718980607, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 123015.65625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 354087895040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29876.677734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.7257537841797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.71392822265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 90763.1015625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2145.5068359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1613.238525390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6363.23876953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4126.1640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29876.677734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.69137573242188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 112.71397399902344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 90762.9375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1164.3514772477822, + "step": 63, + "teaching_rms": 1164.3514446415068, + "traffic_rms": 0.11165118469247219 + }, + { + "batch_loss": 59.3968620300293, + "innovation_rms": 10116327.153258527, + "instruction_rms": 10116327.153258527, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 110749.53125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 318679941120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26879.93359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 145.86865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.42896270751953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 93451.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2285.534423828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1518.615966796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 6734.59375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3713.547607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26879.93359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 145.8342742919922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 101.42900085449219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 93450.9296875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10116327.153258527, + "step": 64, + "teaching_rms": 10116327.153258527, + "traffic_rms": 0.10297234087167896 + }, + { + "batch_loss": 716.4456787109375, + "innovation_rms": 95692.92997692744, + "instruction_rms": 95692.92997688858, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 99674.578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 286811947008.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24182.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.94789123535156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.79237365722656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 95869.3515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2411.55908203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1366.75439453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7068.81298828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3342.19287109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24182.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 154.91351318359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.79241180419922, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 95869.1875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 95692.92997694827, + "step": 65, + "teaching_rms": 95692.92997692744, + "traffic_rms": 0.10737248756990989 + }, + { + "batch_loss": 205.44073486328125, + "innovation_rms": 761030.2552518495, + "instruction_rms": 761030.2552518464, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 89707.1171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 258130755584.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36857.796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.10960388183594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.61711120605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 98044.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2524.9814453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230.0789794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7369.6103515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3007.973388671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36857.796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 163.07522583007812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.61714935302734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 98044.6640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 761030.2552518513, + "step": 66, + "teaching_rms": 761030.2552518495, + "traffic_rms": 0.08801240725381025 + }, + { + "batch_loss": 247.6087188720703, + "innovation_rms": 149018.1374245758, + "instruction_rms": 149018.1374245557, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 80736.40625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 232317681664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33171.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.44300842285156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.33399963378906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 100001.7734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2627.0615234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1188.77880859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7640.328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2707.176025390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33171.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.40863037109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.33403015136719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 100001.609375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 149018.1374245838, + "step": 67, + "teaching_rms": 149018.1374245758, + "traffic_rms": 0.09170831181844329 + }, + { + "batch_loss": 373.77630615234375, + "innovation_rms": 631615.7781113124, + "instruction_rms": 631615.778111304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 72662.765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 209085906944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29853.78125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.99472045898438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.51718139648438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101762.0234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2718.93359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1245.2667236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 7883.97412109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2436.458251953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29853.78125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 176.96035766601562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.5172119140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 101761.859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 631615.7781113103, + "step": 68, + "teaching_rms": 631615.7781113124, + "traffic_rms": 0.08869441008062406 + }, + { + "batch_loss": 43.323360443115234, + "innovation_rms": 10937.70847418805, + "instruction_rms": 10937.708473406581, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 65396.48828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 188177317888.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26867.404296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.88949584960938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.94776153564453, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103345.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2801.618408203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1371.9970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8103.25537109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2192.812255859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26867.404296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 182.85513305664062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.947792053222656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 103345.0703125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10937.70847372278, + "step": 69, + "teaching_rms": 10937.70847418805, + "traffic_rms": 0.08322259161864114 + }, + { + "batch_loss": 45.68955612182617, + "innovation_rms": 35572772.48210096, + "instruction_rms": 35572772.48210096, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 67072.171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 198735429632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24179.396484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.19296264648438, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.03469467163086, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104769.0859375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2876.03466796875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1234.79736328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8300.6083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1973.531005859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24179.396484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 188.15859985351562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.03472900390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 104768.921875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 35572772.48210096, + "step": 70, + "teaching_rms": 35572772.48210096, + "traffic_rms": 0.06997957217143592 + }, + { + "batch_loss": 72.63777160644531, + "innovation_rms": 48961.191689618805, + "instruction_rms": 48961.19168871419, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 60390.96484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 178862030848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21759.947265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.96420288085938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.712406158447266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106049.5078125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 2943.00927734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1117.19091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8478.2265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1776.1778564453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21759.947265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 192.92984008789062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.712440490722656, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 106049.34375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 48961.1916889527, + "step": 71, + "teaching_rms": 48961.191689618805, + "traffic_rms": 0.15454000084123007 + }, + { + "batch_loss": 51.907379150390625, + "innovation_rms": 7365.387748654858, + "instruction_rms": 7365.387747330699, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 54627.41796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 160979025920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19582.224609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.2563934326172, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.921871185302734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107200.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3003.28662109375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1091.9713134765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8638.0830078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1598.56005859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19582.224609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.22203063964844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.92190170288086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 107200.6640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7365.387747971434, + "step": 72, + "teaching_rms": 7365.387748654858, + "traffic_rms": 0.09762449570484948 + }, + { + "batch_loss": 5949.06689453125, + "innovation_rms": 33146.21152242152, + "instruction_rms": 33146.211414310215, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 50625.01953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 450093711360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147648.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.48843383789062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.08814239501953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108235.9453125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3059.92138671875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13992.5615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8779.6455078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35978.36328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 147648.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.45407104492188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.087989807128906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 108235.78125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33146.21144910159, + "step": 73, + "teaching_rms": 33146.21152242152, + "traffic_rms": 1.5196473861236441 + }, + { + "batch_loss": 7292.31591796875, + "innovation_rms": 7364546861.849872, + "instruction_rms": 7364546861.849872, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 473977.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3563353800704.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132881.953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.842041015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.57435607910156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109166.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 3392.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23140.388671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 8907.0517578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26096.916015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 132881.953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.80767822265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.574344635009766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 109166.3046875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7364546861.849872, + "step": 74, + "teaching_rms": 7364546861.849872, + "traffic_rms": 12.84343339413785 + }, + { + "batch_loss": 259.0882873535156, + "innovation_rms": 1152140.185484792, + "instruction_rms": 1152140.1854803285, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 821333.125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5315254288384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119590.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.1582794189453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.87348937988281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110002.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 5351.22802734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30042.3515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9021.716796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23416.017578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 119590.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.12393188476562, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.873477935791016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110002.6796875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1152140.185538239, + "step": 75, + "teaching_rms": 1152140.185484792, + "traffic_rms": 11.551734564839107 + }, + { + "batch_loss": 840.32421875, + "innovation_rms": 82.05639928977297, + "instruction_rms": 81.88856259601224, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 772696.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 14719875284992.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220511.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.42019653320312, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.070072174072266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110754.484375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 8522.5400390625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31713.115234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9124.916015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20570.3203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 220511.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.38584899902344, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.070064544677734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 110754.3203125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 82.02821089438132, + "step": 76, + "teaching_rms": 82.05639928977297, + "traffic_rms": 4.784475384170825 + }, + { + "batch_loss": 1997.2177734375, + "innovation_rms": 208858440.19332775, + "instruction_rms": 208858440.19332775, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 695426.6875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13247887441920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198458.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.41690063476562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.338470458984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111429.8515625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 11376.720703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28541.802734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9217.794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18513.287109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 198458.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.38255310058594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.33846664428711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 111429.6875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 208858440.19332775, + "step": 77, + "teaching_rms": 208858440.19332775, + "traffic_rms": 0.14988704158466135 + }, + { + "batch_loss": 4261.8828125, + "innovation_rms": 28913118.469627153, + "instruction_rms": 28913118.469627146, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 625884.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11923098697728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178608.109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.00485229492188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.30683898925781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112036.5703125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 13945.482421875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25687.62109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9301.3857421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16661.95703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 178608.109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.9705047607422, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.30693054199219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112036.40625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28913118.46962715, + "step": 78, + "teaching_rms": 28913118.469627153, + "traffic_rms": 0.19705822829599642 + }, + { + "batch_loss": 1241.7724609375, + "innovation_rms": 7600604.911146745, + "instruction_rms": 7600604.911146743, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 563295.625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10730789142528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160741.328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.6320037841797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.23587799072266, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112581.4921875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 16257.3681640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23118.857421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9376.6171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14995.7607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160741.328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.59765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.23606872558594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 112581.328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 7600604.911146744, + "step": 79, + "teaching_rms": 7600604.911146745, + "traffic_rms": 0.08673790305502886 + }, + { + "batch_loss": 1902.4310302734375, + "innovation_rms": 2883219.6745009255, + "instruction_rms": 2883219.6745009176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 506966.0625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9657710018560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144659.609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.13165283203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.792877197265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113070.796875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 18338.064453125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20806.970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9444.326171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13496.1845703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 144659.609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.09732055664062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 57.79306411743164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113070.6328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2883219.6745009194, + "step": 80, + "teaching_rms": 2883219.6745009255, + "traffic_rms": 0.09416903355957683 + }, + { + "batch_loss": 844.3853759765625, + "innovation_rms": 266091.74697477254, + "instruction_rms": 266091.7469746488, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 456269.46875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691939016704.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130184.6171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.6126708984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.333309173583984, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113510.0390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 20210.69140625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18726.2734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9505.263671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12146.5654296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130184.6171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.57833862304688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 51.33349609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 113509.875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 266091.7469746452, + "step": 81, + "teaching_rms": 266091.74697477254, + "traffic_rms": 0.09631282162521156 + }, + { + "batch_loss": 1558.805419921875, + "innovation_rms": 1599338.45800607, + "instruction_rms": 1599338.4580060672, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 410642.53125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7822745010176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117155.828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.1436004638672, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.66987609863281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115032.328125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 21896.056640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16853.646484375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9560.107421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10931.908203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 117155.828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 198.10926818847656, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.66976547241211, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 115032.359375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1599338.4580060686, + "step": 82, + "teaching_rms": 1599338.45800607, + "traffic_rms": 0.06482730943630734 + }, + { + "batch_loss": 3305.1689453125, + "innovation_rms": 75001.28440809395, + "instruction_rms": 75001.28440800995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 369578.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7040470614016.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105428.7421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.48373413085938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.49239349365234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 125575.203125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 23412.884765625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15168.28125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 9609.466796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9838.716796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 105428.7421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.4494171142578, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 114.49227905273438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 125575.234375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 75001.2844080289, + "step": 83, + "teaching_rms": 75001.28440809395, + "traffic_rms": 0.054547511758923764 + }, + { + "batch_loss": 672.3721923828125, + "innovation_rms": 265932.49983023526, + "instruction_rms": 265932.49983020505, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 332620.4375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6336423657472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94873.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.88787841796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.38472747802734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 135062.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 24778.029296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13651.453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 10275.8466796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8854.8447265625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 94873.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.85357666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 99.38461303710938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 135062.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 265932.4998302202, + "step": 84, + "teaching_rms": 265932.49983023526, + "traffic_rms": 0.0908454686000366 + }, + { + "batch_loss": 2271.1162109375, + "innovation_rms": 4325618.501894768, + "instruction_rms": 4325618.5018947655, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 299358.40625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5702781239296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85372.46875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.32516479492188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48123168945312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 143599.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 26006.66015625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12286.3076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11072.7822265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7969.35986328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 85372.46875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.29086303710938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48112487792969, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 143599.8125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4325618.501894766, + "step": 85, + "teaching_rms": 4325618.501894768, + "traffic_rms": 0.06675792938215408 + }, + { + "batch_loss": 152.3937530517578, + "innovation_rms": 1856.59201868768, + "instruction_rms": 1856.592017309222, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 269422.5625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5132503220224.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76820.859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.81675720214844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.79853820800781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 151281.875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 27112.427734375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11057.6767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 11790.0244140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7172.423828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 76820.859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.78245544433594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.79843139648438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 151281.90625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1856.5920181130946, + "step": 86, + "teaching_rms": 1856.59201868768, + "traffic_rms": 0.05401767380816191 + }, + { + "batch_loss": 327.6507568359375, + "innovation_rms": 24135.696659637317, + "instruction_rms": 24135.696659599616, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 242480.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4619253055488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69123.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.35723876953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.16736602783203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158194.234375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 28107.619140625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9951.9091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 12435.54296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6455.18115234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69123.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.32293701171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.1672592163086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 158194.265625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 24135.69665967624, + "step": 87, + "teaching_rms": 24135.696659637317, + "traffic_rms": 0.04325624685433459 + }, + { + "batch_loss": 203.13204956054688, + "innovation_rms": 2869.9443332288215, + "instruction_rms": 2869.944332846345, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 218232.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4157327802368.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62195.4609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.94171142578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.48736572265625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 164413.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29003.291015625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8956.7177734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13016.509765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5809.6630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 62195.4609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.90740966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.48725891113281, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 164413.8125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2869.9443330659496, + "step": 88, + "teaching_rms": 2869.9443332288215, + "traffic_rms": 0.03428959260955684 + }, + { + "batch_loss": 39.23117446899414, + "innovation_rms": 735.585376497103, + "instruction_rms": 735.5853756846246, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 196409.046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3741594943488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55959.47265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.5657958984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.63496780395508, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 170009.734375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 29809.396484375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8061.0458984375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 13539.3798828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5228.69677734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55959.47265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.531494140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.63486099243164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 170009.765625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 735.5853761796513, + "step": 89, + "teaching_rms": 735.585376497103, + "traffic_rms": 0.026819681747595088 + }, + { + "batch_loss": 14.122994422912598, + "innovation_rms": 336.7656642176396, + "instruction_rms": 336.7656633965037, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 176768.140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3367435501568.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50346.5234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.22552490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.367286682128906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 175044.390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 30534.890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7254.94091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14009.962890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4705.8271484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 50346.5234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 194.19122314453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.3671760559082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 175044.421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 336.7656640594655, + "step": 90, + "teaching_rms": 336.7656642176396, + "traffic_rms": 0.020985345841795084 + }, + { + "batch_loss": 17.963912963867188, + "innovation_rms": 3172.4900378929187, + "instruction_rms": 3172.490037850651, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 159091.328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3030691872768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45294.3671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.91734313964844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.62590026855469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 179573.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31187.8359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6529.44677734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14433.4873046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4235.244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 45294.3671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.88304138183594, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.625789642333984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 179573.859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3172.490037892819, + "step": 91, + "teaching_rms": 3172.4900378929187, + "traffic_rms": 0.016208204895512557 + }, + { + "batch_loss": 34798.5078125, + "innovation_rms": 56.21518078935135, + "instruction_rms": 55.73845907678787, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3799450.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2636418966880256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103991.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.63803100585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.35667419433594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 183648.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 31775.486328125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5876.501953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 14787.1767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24900.791015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 103991.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.60372924804688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.35675048828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 183648.5625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 55.84365332084459, + "step": 92, + "teaching_rms": 56.21518078935135, + "traffic_rms": 3.4276971823622397 + }, + { + "batch_loss": 1450.1468505859375, + "innovation_rms": 55.555337740404646, + "instruction_rms": 55.53033899674401, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3813598.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5335226091831296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93591.4375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.43551635742188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 224.07655334472656, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 187314.390625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32308.556640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5330.7109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15097.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22125.76953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93591.4375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.40122985839844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 224.0766143798828, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 187314.421875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 55.543188036364036, + "step": 93, + "teaching_rms": 55.555337740404646, + "traffic_rms": 1.1943028390434307 + }, + { + "batch_loss": 5330.20166015625, + "innovation_rms": 2014177.681224092, + "instruction_rms": 2014177.6812233496, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3432238.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4801703643709440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1584325.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.25132751464844, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.12664794921875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 190611.796875, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 32793.953125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4853.951171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15303.646484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19913.19140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1584325.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 193.217041015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 195.126708984375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 190611.828125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2014177.681223577, + "step": 94, + "teaching_rms": 2014177.681224092, + "traffic_rms": 0.9577668411703144 + }, + { + "batch_loss": 13750.9228515625, + "innovation_rms": 369644.0973330957, + "instruction_rms": 369644.09733159625, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3089014.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4321533279338496.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2592381.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.62945556640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.9471893310547, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 301019.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33230.80859375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4368.55615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15488.9541015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17921.87109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2592381.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 190.59518432617188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 172.94703674316406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 301019.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 369644.0973322402, + "step": 95, + "teaching_rms": 369644.0973330957, + "traffic_rms": 0.6940779991497632 + }, + { + "batch_loss": 4439.3671875, + "innovation_rms": 1629.6140078430903, + "instruction_rms": 1629.609929165954, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11858387.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8349087321554944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2333117.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.97500610351562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.18533325195312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 492549.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33623.98046875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10156.64453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15604.1533203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29920.73046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2333117.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 187.9407501220703, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 201.18511962890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 492549.6875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1629.620637637452, + "step": 96, + "teaching_rms": 1629.6140078430903, + "traffic_rms": 5.908032157801106 + }, + { + "batch_loss": 39616.0625, + "innovation_rms": 4486436.411854939, + "instruction_rms": 4486436.41185492, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10672548.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7514178428338176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2099756.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.392333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 641.773681640625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 702525.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 33977.83203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20556.421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15674.3740234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26928.65625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2099756.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.3580780029297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 641.7738037109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 702525.3125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4486436.411854932, + "step": 97, + "teaching_rms": 4486436.411854939, + "traffic_rms": 0.3419906995368841 + }, + { + "batch_loss": 29161.96875, + "innovation_rms": 154396942.94943586, + "instruction_rms": 154396942.94943586, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9605293.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6762760478130176.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889710.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.8644256591797, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 540.0814208984375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 891496.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34296.30078125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18500.779296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15737.572265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24235.7890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1889710.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 162.83016967773438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 540.08154296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 891496.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 154396942.94943586, + "step": 98, + "teaching_rms": 154396942.94943586, + "traffic_rms": 0.19069054955236614 + }, + { + "batch_loss": 72121.9921875, + "innovation_rms": 3874518777.135761, + "instruction_rms": 3874518777.135761, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8644764.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6086484591378432.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1700650.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.27090454101562, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.2171630859375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1061561.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34582.921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16650.701171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15794.451171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21812.208984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1700650.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 138.23666381835938, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 818.2172241210938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1061561.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3874518777.135761, + "step": 99, + "teaching_rms": 3874518777.135761, + "traffic_rms": 0.12258361498714034 + }, + { + "batch_loss": 26738.458984375, + "innovation_rms": 1705363243.6050854, + "instruction_rms": 1705363243.6050854, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7780287.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5477836185927680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530479.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.6424102783203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 705.591796875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1214609.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 34840.87890625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89920.078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15845.642578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19630.98828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530479.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 144.48326110839844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 705.5918579101562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1214609.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1705363243.6050854, + "step": 100, + "teaching_rms": 1705363243.6050854, + "traffic_rms": 0.23018973295536913 + }, + { + "batch_loss": 21163.857421875, + "innovation_rms": 45894576.22109617, + "instruction_rms": 45894576.22109617, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7002258.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4930052298899456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1377309.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 171.15792846679688, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 612.1289672851562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1352340.25, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 90535.5234375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 905355.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15891.71484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17667.888671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1377309.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.998779296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 612.1290283203125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1352340.25, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 45894576.22109617, + "step": 101, + "teaching_rms": 45894576.22109617, + "traffic_rms": 0.21639849755658833 + }, + { + "batch_loss": 2820.987060546875, + "innovation_rms": 17.1334560491678, + "instruction_rms": 2.1544434067781384, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 89562480.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1193850255701443e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1239443.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.03269958496094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 548.3017578125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1476284.625, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 166117.515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 755819.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15933.1796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15901.099609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1239443.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 199.92633056640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 548.3018188476562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1476284.625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31.296822141660023, + "step": 102, + "teaching_rms": 17.1334560491678, + "traffic_rms": 31.22258063099652 + }, + { + "batch_loss": 190.3369598388672, + "innovation_rms": 31.177775607304177, + "instruction_rms": 0.057912260984576636, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 392813888.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4611736576468713e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1115351.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.3778533935547, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 493.4515686035156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1587819.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 223315.828125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 571983.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 15970.498046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14310.9892578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1115351.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 249.27149963378906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 493.4516296386719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1587819.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 57.2870409060788, + "step": 103, + "teaching_rms": 31.177775607304177, + "traffic_rms": 57.28701508518365 + }, + { + "batch_loss": 21623.69921875, + "innovation_rms": 10349250224.125568, + "instruction_rms": 10349250224.125568, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 952185408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.316665714270798e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1003657.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 293.01690673828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 436.39056396484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1688185.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 260241.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 369255.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16004.083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12879.8896484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1003657.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 292.9105529785156, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 436.3905944824219, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1688185.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10349250224.125568, + "step": 104, + "teaching_rms": 10349250224.125568, + "traffic_rms": 77.02182068991524 + }, + { + "batch_loss": 654.7659301757812, + "innovation_rms": 1160356.4762901121, + "instruction_rms": 1160356.4752604428, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1742342528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6622969183359363e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 903122.8125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 332.28912353515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 392.7221984863281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1778497.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 276514.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 162736.390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16034.3115234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18680.341796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 903122.8125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 332.1827697753906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 392.72222900390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1778497.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1160356.4787314876, + "step": 105, + "teaching_rms": 1160356.4762901121, + "traffic_rms": 89.75131591670393 + }, + { + "batch_loss": 28.654651641845703, + "innovation_rms": 614759.5286315297, + "instruction_rms": 614759.5264401266, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2688386560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.246996815092279e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 812632.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 367.63079833984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 353.416748046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859761.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 273141.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47787.203125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 16281.65234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24178.984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 812632.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 367.5244445800781, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 353.4167785644531, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859761.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 614759.5338360644, + "step": 106, + "teaching_rms": 614759.5286315297, + "traffic_rms": 95.3595642139492 + }, + { + "batch_loss": 67.26221466064453, + "innovation_rms": 992347.4808665419, + "instruction_rms": 992347.4795422661, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3685697792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.158950075084057e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 731183.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 399.43463134765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 318.0382995605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1932879.375, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 252305.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 208361.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 17220.595703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26787.521484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 731183.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 399.3282775878906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 318.0383605957031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1932879.375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 992347.4840128833, + "step": 107, + "teaching_rms": 992347.4808665419, + "traffic_rms": 94.19560326182284 + }, + { + "batch_loss": 22.603679656982422, + "innovation_rms": 817722.773814597, + "instruction_rms": 817722.772443721, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4619218432.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.00868962466092e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 657871.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 428.0540771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286.19451904296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1998666.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 217110.515625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 351947.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18065.64453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24108.76953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 657871.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 427.9477233886719, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 286.194580078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1998666.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 817722.7770727874, + "step": 108, + "teaching_rms": 817722.773814597, + "traffic_rms": 87.00911365584236 + }, + { + "batch_loss": 35.365901947021484, + "innovation_rms": 188064.4477042057, + "instruction_rms": 188064.44329043356, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5380583424.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.395158658006458e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 591884.5625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 453.80731201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257.5322570800781, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2057855.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 171286.703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 458238.125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 18826.189453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21697.892578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 591884.5625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 453.7009582519531, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 257.5323181152344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2057855.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 188064.45819455187, + "step": 109, + "teaching_rms": 188064.4477042057, + "traffic_rms": 74.87235363678894 + }, + { + "batch_loss": 76704.328125, + "innovation_rms": 791882062.3084093, + "instruction_rms": 791882062.3084087, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5880616448.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.0584945508093e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8878180.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 474.169677734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.87893676757812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2111104.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 118882.9296875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 524037.75, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 19510.6796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19528.103515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8878180.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 474.0633239746094, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.87892150878906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2111104.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 791882062.3084109, + "step": 110, + "teaching_rms": 791882062.3084093, + "traffic_rms": 59.068752864613934 + }, + { + "batch_loss": 133819.8125, + "innovation_rms": 3404781711.383887, + "instruction_rms": 3404781711.383887, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6057617408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.970769355880215e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7990273.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 486.6067199707031, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1960.5316162109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2159007.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 64523.9609375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 549107.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20126.720703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17575.29296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7990273.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 486.5003662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1960.53173828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2159007.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3404781711.383887, + "step": 111, + "teaching_rms": 3404781711.383887, + "traffic_rms": 41.0166035170652 + }, + { + "batch_loss": 139979.8125, + "innovation_rms": 21167066894.381527, + "instruction_rms": 21167066894.381527, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5885148672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.336531642482279e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191077.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 494.7956237792969, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1908.21337890625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2405952.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 67063.5703125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 535885.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 20681.158203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15817.763671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7191077.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 494.68927001953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1908.21337890625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 2405953.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 21167066894.381527, + "step": 112, + "teaching_rms": 21167066894.381527, + "traffic_rms": 22.143065772060186 + }, + { + "batch_loss": 1747.673095703125, + "innovation_rms": 677463190.7983509, + "instruction_rms": 677463190.7983509, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5369628672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.510435659146267e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6471728.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 502.1606750488281, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1717.375244140625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3053125.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 69349.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 489063.375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21180.15234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14235.9873046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6471728.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 502.0543212890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1717.375244140625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3053125.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 677463190.7983509, + "step": 113, + "teaching_rms": 677463190.7983509, + "traffic_rms": 4.032699681702085 + }, + { + "batch_loss": 1819.5068359375, + "innovation_rms": 3681439843060.2715, + "instruction_rms": 3681439843060.2715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4832665600.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.759392163600385e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5824250.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 572.256103515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1545.603759765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3635550.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 82538.359375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 440157.03125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 21629.24609375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12812.388671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5824250.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 572.0428466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1545.603759765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 3635550.75, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3681439843060.2715, + "step": 114, + "teaching_rms": 3681439843060.2715, + "traffic_rms": 1.4512551612798417 + }, + { + "batch_loss": 9276.65234375, + "innovation_rms": 456866884546827.94, + "instruction_rms": 456866884546827.94, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4349399040.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.083453228715323e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5241461.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 676.1685180664062, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1390.9940185546875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4159696.75, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 122152.4921875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 396141.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22033.4296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11531.1494140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5241461.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 675.9552612304688, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1390.9940185546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4159697.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 456866884546827.94, + "step": 115, + "teaching_rms": 456866884546827.94, + "traffic_rms": 0.8427960906421251 + }, + { + "batch_loss": 7904.04150390625, + "innovation_rms": 1492190030155.8389, + "instruction_rms": 1492190030155.8389, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3914457344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.475108046581279e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4716899.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.6829223632812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1251.831298828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4631386.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 157805.203125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 356527.15625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22397.1953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10378.0341796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4716899.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 769.4696655273438, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1251.831298828125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 4631387.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1492190030155.8389, + "step": 116, + "teaching_rms": 1492190030155.8389, + "traffic_rms": 0.34813901344231035 + }, + { + "batch_loss": 40475.6640625, + "innovation_rms": 106669291.61713824, + "instruction_rms": 106669291.61713824, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3523007232.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.9275973474762673e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4244746.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 870.6068725585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1126.5723876953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5055861.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 189892.640625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 320874.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 22724.583984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9340.23046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4244746.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 870.6082763671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1126.5723876953125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5055861.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 106669291.61713824, + "step": 117, + "teaching_rms": 106669291.61713824, + "traffic_rms": 0.3382242943798017 + }, + { + "batch_loss": 11094.6171875, + "innovation_rms": 64551.29359661899, + "instruction_rms": 64551.29359557233, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3170694400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.434837753466129e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3819765.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 971.9896850585938, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1013.8280639648438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5437837.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 218771.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288787.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23019.234375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8406.20703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3819765.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 971.9910888671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1013.8280639648438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5437838.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 64551.29359614055, + "step": 118, + "teaching_rms": 64551.29359661899, + "traffic_rms": 0.2715866267992134 + }, + { + "batch_loss": 3310.7578125, + "innovation_rms": 23209.36942110678, + "instruction_rms": 23209.369419991126, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2853601536.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9913538725664e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3437245.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063.2244873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 912.3480224609375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5781562.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 244762.171875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 259908.296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23284.419921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7565.5859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3437245.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1063.2259521484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 912.3480224609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 5781562.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23209.36942155309, + "step": 119, + "teaching_rms": 23209.36942110678, + "traffic_rms": 0.2703935283034405 + }, + { + "batch_loss": 2535.720703125, + "innovation_rms": 67877.8695883476, + "instruction_rms": 67877.86958775036, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2568225792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.59221848530976e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3092942.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1145.3251953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 821.0068969726562, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6090856.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 268153.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 233917.46875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23523.0859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6809.02734375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3092942.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1145.32666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 821.0068969726562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6090857.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 67877.86958831549, + "step": 120, + "teaching_rms": 67877.8695883476, + "traffic_rms": 0.27581190291462626 + }, + { + "batch_loss": 445.22064208984375, + "innovation_rms": 13070.20080237828, + "instruction_rms": 13070.200801421683, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2311390208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.232996601594412e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2783039.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1219.204345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 738.7916259765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6369160.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 289206.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210525.71875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23737.88671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6128.12451171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2783039.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1219.205810546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 738.7916259765625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6369161.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 13070.200803096188, + "step": 121, + "teaching_rms": 13070.20080237828, + "traffic_rms": 0.2092501311182215 + }, + { + "batch_loss": 4145.38037109375, + "innovation_rms": 3321861.5081997146, + "instruction_rms": 3321861.5081997016, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2080240128.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9096969062505985e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2504098.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1285.683349609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 664.79052734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6619570.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 308153.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 189473.140625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 23931.20703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5515.31201171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2504098.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1285.684814453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 664.79052734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6619571.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3321861.5081997057, + "step": 122, + "teaching_rms": 3321861.5081997146, + "traffic_rms": 0.18719239052925485 + }, + { + "batch_loss": 30927.05859375, + "innovation_rms": 27850535.482121, + "instruction_rms": 27850535.48212099, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872207744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6187272508099107e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2253026.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1345.5015869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 598.1829223632812, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6844873.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 325206.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170525.828125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24105.1953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4963.78076171875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2253026.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1345.5030517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 598.1829223632812, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 6844873.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 27850535.48212099, + "step": 123, + "teaching_rms": 27850535.482121, + "traffic_rms": 0.21867303397486074 + }, + { + "batch_loss": 23206.380859375, + "innovation_rms": 2956950822.285266, + "instruction_rms": 2956950822.285266, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1684978688.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3568545433211057e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2027039.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1399.3245849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 538.2301025390625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7047577.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 340553.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153473.234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24261.78515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4503.40625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2027039.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1399.3260498046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 538.2301025390625, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7047577.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2956950822.285266, + "step": 124, + "teaching_rms": 2956950822.285266, + "traffic_rms": 0.20347581446332308 + }, + { + "batch_loss": 49260.859375, + "innovation_rms": 3116872821.463919, + "instruction_rms": 3116872821.463919, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1516471936.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.121169036212437e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1823630.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1447.7513427734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 532.6173095703125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7229940.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 354366.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 138125.90625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24402.71484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4181.15966796875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1823630.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1447.7528076171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 532.6171264648438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7229940.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3116872821.463919, + "step": 125, + "teaching_rms": 3116872821.463919, + "traffic_rms": 0.16762426641332961 + }, + { + "batch_loss": 37775.8125, + "innovation_rms": 3726014308.5505114, + "instruction_rms": 3726014308.5505114, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1364815872.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9090521325911933e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1640544.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1491.3209228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 444.3330078125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7393994.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 366797.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124313.3125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24529.552734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3618.595703125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1640544.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1491.3223876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 444.33282470703125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7393995.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3726014308.5505114, + "step": 126, + "teaching_rms": 3726014308.5505114, + "traffic_rms": 0.1545241170647442 + }, + { + "batch_loss": 22684.005859375, + "innovation_rms": 912923143.828508, + "instruction_rms": 912923143.828508, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1228326016.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7181468313711437e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1475750.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530.5186767578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 391.9769287109375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7541569.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 377985.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111881.9765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24643.70703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3256.736083984375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1475750.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1530.5201416015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 391.9769287109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7541570.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 912923143.828508, + "step": 127, + "teaching_rms": 912923143.828508, + "traffic_rms": 0.190676409287617 + }, + { + "batch_loss": 11248.2392578125, + "innovation_rms": 844098271.448619, + "instruction_rms": 844098271.448619, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1105486208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5463322186027735e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1327421.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1565.78125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 352.6261901855469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7674311.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 388055.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100693.7734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24746.4453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2931.0625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1327421.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1565.78271484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 352.6261901855469, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7674312.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 844098271.448619, + "step": 128, + "teaching_rms": 844098271.448619, + "traffic_rms": 0.20476248185497917 + }, + { + "batch_loss": 6286.47802734375, + "innovation_rms": 1530695490.1888525, + "instruction_rms": 1530695490.1888525, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 994931840.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3916990143346822e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193911.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1597.501953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 317.20697021484375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7793702.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 397117.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90624.390625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24838.91015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2637.956298828125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1193911.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1597.50341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 317.20697021484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7793703.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1530695490.1888525, + "step": 129, + "teaching_rms": 1530695490.1888525, + "traffic_rms": 0.21808496229076418 + }, + { + "batch_loss": 4260.3583984375, + "innovation_rms": 1937043802.3431761, + "instruction_rms": 1937043802.3431761, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 895434560.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.252529077716842e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1073741.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1626.0345458984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.3265075683594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7901076.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 405273.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81561.953125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24922.12890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2374.16064453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1073741.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1626.0360107421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 285.3265075683594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7901077.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1937043802.3431761, + "step": 130, + "teaching_rms": 1937043802.3431761, + "traffic_rms": 0.21713706529412918 + }, + { + "batch_loss": 6951.6337890625, + "innovation_rms": 3142017317.2697797, + "instruction_rms": 3142017317.2697797, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 805888768.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1272761435568787e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 965576.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1651.6976318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256.6312561035156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7997634.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 412614.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73405.7578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 24997.025390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2136.74462890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 965576.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1651.6990966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 256.6312561035156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 7997634.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3142017317.2697797, + "step": 131, + "teaching_rms": 3142017317.2697797, + "traffic_rms": 0.19990978667274367 + }, + { + "batch_loss": 9759.3134765625, + "innovation_rms": 18161820896.946182, + "instruction_rms": 18161820896.946182, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 725299328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0145484940168187e+20, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868219.3125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1674.7779541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.80294799804688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8084456.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 419220.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 66065.1796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25064.431640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1923.070068359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 868219.3125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1674.7794189453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.80294799804688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8084456.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 18161820896.946182, + "step": 132, + "teaching_rms": 18161820896.946182, + "traffic_rms": 0.17481659425787535 + }, + { + "batch_loss": 17075.978515625, + "innovation_rms": 245947640.18787143, + "instruction_rms": 245947640.18787143, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 652769408.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.130936534112299e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 780588.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1695.533447265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.55516052246094, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8162515.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 425166.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59458.66015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25125.09765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730.7630615234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 780588.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1695.534912109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.55516052246094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8162515.5, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 245947640.18787143, + "step": 133, + "teaching_rms": 245947640.18787143, + "traffic_rms": 0.20005679938945128 + }, + { + "batch_loss": 157.33346557617188, + "innovation_rms": 1878873.1755825013, + "instruction_rms": 1878873.1755824971, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 587492480.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.21784314458386e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701713.75, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1714.1964111328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.6300811767578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8232686.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 430518.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53512.79296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25179.697265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1557.686767578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 701713.75, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1714.1978759765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.6300811767578, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8232687.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1878873.1755825072, + "step": 134, + "teaching_rms": 1878873.1755825013, + "traffic_rms": 0.19276152983768027 + }, + { + "batch_loss": 626.1778564453125, + "innovation_rms": 28219708.03182532, + "instruction_rms": 28219708.03182532, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 528743232.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.396058478281753e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 630719.125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730.9759521484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.795654296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8295758.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 435334.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48161.51171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25228.8359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1401.9180908203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 630719.125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1730.9774169921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.795654296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8295759.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 28219708.03182532, + "step": 135, + "teaching_rms": 28219708.03182532, + "traffic_rms": 0.17212384620494095 + }, + { + "batch_loss": 79797.4609375, + "innovation_rms": 23822876.746210117, + "instruction_rms": 23822876.746210117, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 475868896.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.656452630453577e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 566817.625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1746.060302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311.909423828125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8352440.5, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 439668.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43345.359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25273.060546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1261.7261962890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 566817.625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1746.061767578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 311.90960693359375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8352441.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 23822876.746210117, + "step": 136, + "teaching_rms": 23822876.746210117, + "traffic_rms": 0.1044686088665963 + }, + { + "batch_loss": 107035.4140625, + "innovation_rms": 56.24663941904081, + "instruction_rms": 56.24183297400059, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 428282016.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.990807411388685e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 509300.59375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1759.61865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.5791015625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8403371.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 443569.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 39010.82421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25312.86328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1135.5535888671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 509300.59375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1759.6201171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 247.57928466796875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8403371.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 56.25104374635909, + "step": 137, + "teaching_rms": 56.24663941904081, + "traffic_rms": 1.0181515222660913 + }, + { + "batch_loss": 87151.8203125, + "innovation_rms": 41673.73587703515, + "instruction_rms": 41673.735856783984, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 385453824.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.391726626269351e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 457530.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1771.8035888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.0955047607422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8449124.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 447080.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35109.7421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25348.685546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1235.1380615234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 457530.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1771.8050537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.09568786621094, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8449124.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 41673.73589848967, + "step": 138, + "teaching_rms": 41673.73587703515, + "traffic_rms": 1.8645443710527563 + }, + { + "batch_loss": 74592.7109375, + "innovation_rms": 22658.386580346014, + "instruction_rms": 22658.385531534484, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 346908448.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.852553831701021e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410932.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1782.7523193359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 326.108154296875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490217.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 450240.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31598.767578125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25380.92578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1471.967529296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410932.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1782.7537841796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 326.108154296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490217.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22658.38567014104, + "step": 139, + "teaching_rms": 22658.386580346014, + "traffic_rms": 2.5062198463671486 + }, + { + "batch_loss": 46942.4921875, + "innovation_rms": 14880.32028520808, + "instruction_rms": 14880.320148654637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 380769280.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.367298448530919e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 368989.96875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1792.58837890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 439.2528076171875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8527116.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 453084.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28438.890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25409.94140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1605.368408203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 368989.96875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1792.58984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 439.2528076171875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8527116.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 14880.320435427042, + "step": 140, + "teaching_rms": 14880.32028520808, + "traffic_rms": 2.9213298661258715 + }, + { + "batch_loss": 28753.6875, + "innovation_rms": 12001.57551193733, + "instruction_rms": 12001.57532106304, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 587518208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9305684717364314e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 331238.25, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1801.4228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384.9183044433594, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560240.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 455644.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25595.0, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25436.056640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1666.0244140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 331238.25, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1801.42431640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 384.9183044433594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8560240.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12001.575722879803, + "step": 141, + "teaching_rms": 12001.57551193733, + "traffic_rms": 3.1054520062815887 + }, + { + "batch_loss": 16195.46875, + "innovation_rms": 109775.9651925522, + "instruction_rms": 109775.96517218684, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 805671872.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.537511536601858e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 297258.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809.35595703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 340.9581604003906, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8589966.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 457947.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23035.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25459.560546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1663.9765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 297258.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1809.357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 340.95819091796875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8589966.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 109775.96521511381, + "step": 142, + "teaching_rms": 109775.9651925522, + "traffic_rms": 3.0696885278088266 + }, + { + "batch_loss": 7430.72607421875, + "innovation_rms": 93113.69734236189, + "instruction_rms": 93113.69732184726, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1010178176.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1837603829416722e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266673.53125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1816.4776611328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303.6713562011719, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8616633.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 460020.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20731.94921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25480.712890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1566.0447998046875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266673.53125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1816.4791259765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 303.67138671875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8616633.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 93113.69736510878, + "step": 143, + "teaching_rms": 93113.69734236189, + "traffic_rms": 2.8383274457601986 + }, + { + "batch_loss": 4553.58544921875, + "innovation_rms": 33081.90680798647, + "instruction_rms": 33081.906765117565, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1177340160.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.865384256686575e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239144.515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1822.8690185546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271.5569152832031, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8640547.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 461886.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18856.265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25499.75, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1457.2144775390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239144.515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1822.8704833984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 271.55694580078125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8640547.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 33081.906855546156, + "step": 144, + "teaching_rms": 33081.90680798647, + "traffic_rms": 2.445925048436639 + }, + { + "batch_loss": 1524.624267578125, + "innovation_rms": 22796.16289090778, + "instruction_rms": 22796.162852018784, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1287683712.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5788458090276848e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214366.0, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1828.60302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 243.92755126953125, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8661984.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 463565.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21594.22265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25516.884765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1332.101318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 214366.0, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1828.6044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 243.92758178710938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8661984.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 22796.162934073775, + "step": 145, + "teaching_rms": 22796.16289090778, + "traffic_rms": 1.934041507024765 + }, + { + "batch_loss": 195.80589294433594, + "innovation_rms": 25777.57112942025, + "instruction_rms": 25777.571112710106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1327911552.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.320961294095614e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192063.1875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833.745361328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.53981018066406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681190.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 465077.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22656.904296875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25532.3046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1198.89111328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 192063.1875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1833.746826171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 219.5398406982422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681190.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 25777.571147947478, + "step": 146, + "teaching_rms": 25777.57112942025, + "traffic_rms": 1.3478437539394923 + }, + { + "batch_loss": 159.88479614257812, + "innovation_rms": 9945457.076168535, + "instruction_rms": 9945457.076168522, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1291916672.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.0888651646860526e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171988.734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1838.3551025390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.58865356445312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698389.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 466437.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22142.82421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25546.18359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1079.001953125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 171988.734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1838.3565673828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 197.5886688232422, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698389.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9945457.07616855, + "step": 147, + "teaching_rms": 9945457.076168535, + "traffic_rms": 0.732810979297577 + }, + { + "batch_loss": 142.257080078125, + "innovation_rms": 229021.32828031149, + "instruction_rms": 229021.32828029245, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1180918528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8799786262272147e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153920.015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1842.4854736328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.8306427001953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713781.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 467661.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20243.107421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25558.673828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 971.1017456054688, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153920.015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1842.4869384765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 177.83065795898438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713781.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 229021.32828033034, + "step": 148, + "teaching_rms": 229021.32828031149, + "traffic_rms": 0.13183835471103542 + }, + { + "batch_loss": 510.8158874511719, + "innovation_rms": 130277294.75612995, + "instruction_rms": 130277294.75612995, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1062830848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6919807196240282e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137656.640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.184326171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.04664611816406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727547.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 468763.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18218.796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25569.916015625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 873.9915771484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137656.640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.185791015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 160.04666137695312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727547.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 130277294.75612995, + "step": 149, + "teaching_rms": 130277294.75612995, + "traffic_rms": 0.007074159689247733 + }, + { + "batch_loss": 20248.529296875, + "innovation_rms": 36049338306.00162, + "instruction_rms": 36049338306.00162, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 956613504.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.522782592686044e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123018.21875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.494873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.1591796875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739849.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 469755.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16396.916015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25580.033203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1367.1351318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 123018.21875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.496337890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 230.15931701660156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739849.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 36049338306.00162, + "step": 150, + "teaching_rms": 36049338306.00162, + "traffic_rms": 0.29154960346644127 + }, + { + "batch_loss": 374.1198425292969, + "innovation_rms": 582427223.6219846, + "instruction_rms": 582427223.6219846, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 860953152.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3705043004320907e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109842.40625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.455810546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.14675903320312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750833.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 470647.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14757.2236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25589.138671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230.421630859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109842.40625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.457275390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 207.14691162109375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750833.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 582427223.6219846, + "step": 151, + "teaching_rms": 582427223.6219846, + "traffic_rms": 0.0030013731878210308 + }, + { + "batch_loss": 11800.4111328125, + "innovation_rms": 5926944.797722141, + "instruction_rms": 5926944.797722137, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 774913088.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2334539143693468e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97983.078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.1021728515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.33750915527344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760631.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 471450.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13281.5009765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25597.333984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1107.37939453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 97983.078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.1036376953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 186.33766174316406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760631.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5926944.797722138, + "step": 152, + "teaching_rms": 5926944.797722141, + "traffic_rms": 0.15938694507510712 + }, + { + "batch_loss": 853.9813842773438, + "innovation_rms": 3740.20667231608, + "instruction_rms": 3740.206853122594, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 697462400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1101085009421795e+19, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87308.703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.46533203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.7033233642578, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769362.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 472173.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11953.3505859375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25604.708984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1029.9757080078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 87308.703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 167.70347595214844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769362.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3740.206855394006, + "step": 153, + "teaching_rms": 3740.20667231608, + "traffic_rms": 0.13108184764498002 + }, + { + "batch_loss": 672.5816650390625, + "innovation_rms": 1997.540150362273, + "instruction_rms": 1997.5401425960727, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 627751168.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.99097628857729e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77700.890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.5736083984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.93087768554688, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777132.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 472824.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10758.015625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25611.34765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1315.9783935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 77700.890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.5750732421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 150.9310302734375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777132.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1997.540146139994, + "step": 154, + "teaching_rms": 1997.540150362273, + "traffic_rms": 0.11945034556914633 + }, + { + "batch_loss": 417.1689147949219, + "innovation_rms": 1720.1181991654296, + "instruction_rms": 1720.118192269717, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 565003008.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.991878659719561e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69053.0859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.4525146484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.83416748046875, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784037.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 473409.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9682.2138671875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25617.322265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1513.86181640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 69053.0859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.4539794921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 135.83432006835938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784037.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1720.1181957883275, + "step": 155, + "teaching_rms": 1720.1181991654296, + "traffic_rms": 0.11017883574187924 + }, + { + "batch_loss": 185.9398956298828, + "innovation_rms": 1312.0361824316456, + "instruction_rms": 1312.0361095741766, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 508520704.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.092690903698768e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61269.375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.1248779296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.24578094482422, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790164.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 473936.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713.9921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25622.69921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1629.765869140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61269.375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.1263427734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 122.24593353271484, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790164.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1312.0361135633252, + "step": 156, + "teaching_rms": 1312.0361824316456, + "traffic_rms": 0.10262149527921566 + }, + { + "batch_loss": 59.91779327392578, + "innovation_rms": 16001.35647823779, + "instruction_rms": 16001.356477865094, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 457677792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.28342175835331e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54263.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.611328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.96595764160156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795590.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 474410.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7842.5927734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25627.5390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1668.3740234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 54263.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.61279296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 109.96610260009766, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795590.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 16001.356478150627, + "step": 157, + "teaching_rms": 16001.35647823779, + "traffic_rms": 0.09559555073251423 + }, + { + "batch_loss": 97.5270767211914, + "innovation_rms": 12437.105777768933, + "instruction_rms": 12437.105777443967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 411917344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.555079417591235e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47957.51953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.9305419921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.96207427978516, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800386.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 474837.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7058.33349609375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25631.89453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1636.314453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47957.51953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.9320068359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 98.96221160888672, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800386.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12437.105777768007, + "step": 158, + "teaching_rms": 12437.105777768933, + "traffic_rms": 0.08973776951162989 + }, + { + "batch_loss": 108.66709899902344, + "innovation_rms": 12425.852663317302, + "instruction_rms": 12425.85266306827, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 370733760.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.899571475832111e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42281.7265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.09912109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.05758666992188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804614.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 475221.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6352.5, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25635.814453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1547.53662109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42281.7265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.1005859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 89.05772399902344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804614.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 12425.852663351983, + "step": 159, + "teaching_rms": 12425.852663317302, + "traffic_rms": 0.08387160331231051 + }, + { + "batch_loss": 28.771549224853516, + "innovation_rms": 4743.349480317385, + "instruction_rms": 4743.349773190418, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 333666048.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.309614218297737e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37173.09375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.1322021484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.14265441894531, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808331.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 475567.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5717.25, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25639.341796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1449.739990234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 37173.09375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.1336669921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.14279174804688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808331.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4743.349773712865, + "step": 160, + "teaching_rms": 4743.349480317385, + "traffic_rms": 0.07039108685593892 + }, + { + "batch_loss": 49.8839225769043, + "innovation_rms": 5174.947829060944, + "instruction_rms": 5174.947828824574, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 300302752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.778652906419126e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32574.951171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.0433349609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.11841583251953, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811588.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 475878.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5145.52490234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25642.517578125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1349.797607421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32574.951171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.0447998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.1185531616211, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811588.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 5174.947829111238, + "step": 161, + "teaching_rms": 5174.947829060944, + "traffic_rms": 0.05446490643448948 + }, + { + "batch_loss": 137.26068115234375, + "innovation_rms": 4342.474463375717, + "instruction_rms": 4342.474463230649, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 270274528.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.300787560801632e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28436.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8446044921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.89588165283203, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476158.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4630.97216796875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25645.375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1248.7637939453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28436.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8460693359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 64.89601135253906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4342.474463403391, + "step": 162, + "teaching_rms": 4342.474463375717, + "traffic_rms": 0.03858703743423635 + }, + { + "batch_loss": 5.444925785064697, + "innovation_rms": 8165.902048254391, + "instruction_rms": 8165.902048218558, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 243248400.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.870708914672632e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24711.22265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.547119140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.39495086669922, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816903.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476411.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4167.875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25647.947265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1147.6409912109375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24711.22265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.548583984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 58.395084381103516, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816903.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 8165.902048255262, + "step": 163, + "teaching_rms": 8165.902048254391, + "traffic_rms": 0.024483450354417305 + }, + { + "batch_loss": 3.961790084838867, + "innovation_rms": 1952.2187359412437, + "instruction_rms": 1952.2188249660958, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 218924272.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4836379407419965e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21358.41015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.16064453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.54352951049805, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819039.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476637.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3751.08740234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25650.26171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1047.2542724609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21358.41015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.162109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.54366683959961, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819039.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1952.2188250090103, + "step": 164, + "teaching_rms": 1952.2187359412437, + "traffic_rms": 0.012944329831814956 + }, + { + "batch_loss": 3.811180591583252, + "innovation_rms": 0.265417077280466, + "instruction_rms": 0.26531595657338863, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 197031840.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1352741741555876e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18340.6640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.694091796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.27672576904297, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820873.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 476842.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3375.978515625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25652.345703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 942.52880859375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18340.6640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.695556640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.27686309814453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820873.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26531595657338863, + "step": 165, + "teaching_rms": 0.265417077280466, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.744426727294922, + "innovation_rms": 0.2871283620189236, + "instruction_rms": 0.28708262912720606, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 177328656.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8217467842278195e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15624.5107421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.155517578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.536128997802734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822435.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477025.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3038.380615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25654.220703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 848.27587890625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15624.5107421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.156982421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 42.5362663269043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822435.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28708262912720606, + "step": 166, + "teaching_rms": 0.2871283620189236, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.9801549911499023, + "innovation_rms": 0.2541226041526663, + "instruction_rms": 0.2540026937672959, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 159595792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.539571968366084e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13179.81640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.552001953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.269168853759766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823753.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477191.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2734.54248046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25655.908203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 763.4482421875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13179.81640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.553466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.26930236816406, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823753.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2540026937672959, + "step": 167, + "teaching_rms": 0.2541226041526663, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.7612504959106445, + "innovation_rms": 0.2875131544099313, + "instruction_rms": 0.28745403210754383, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 143636208.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.285614744041685e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10979.458984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.89013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.42851638793945, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824851.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477340.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2461.088134765625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25657.427734375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 687.1033935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10979.458984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.8916015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 34.428653717041016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824851.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28745403210754383, + "step": 168, + "teaching_rms": 0.2875131544099313, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.6806769371032715, + "innovation_rms": 0.3060041205106952, + "instruction_rms": 0.30598188373525176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 129272584.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.057053214661935e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9245.9404296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.17578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.971588134765625, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825751.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477473.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2214.979248046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25658.794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 618.39306640625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9245.9404296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.17724609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 30.97172737121582, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825751.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30598188373525176, + "step": 169, + "teaching_rms": 0.3060041205106952, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.676483154296875, + "innovation_rms": 0.2625952381519611, + "instruction_rms": 0.2624638352844185, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 116345328.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8513479206835323e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8291.470703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.4140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.860042572021484, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826473.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477594.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1993.4813232421875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25660.025390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 556.5537719726562, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8291.470703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.41552734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.86018180847168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826473.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2624638352844185, + "step": 170, + "teaching_rms": 0.2625952381519611, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.4520416259765625, + "innovation_rms": 0.28537347654501183, + "instruction_rms": 0.2852918655692243, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 104710792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6662131148712837e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7432.365234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.60986328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.05937385559082, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827034.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477703.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1794.1331787109375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25661.1328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 500.89837646484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7432.365234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.611328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.059513092041016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827034.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2852918655692243, + "step": 171, + "teaching_rms": 0.28537347654501183, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.229200601577759, + "innovation_rms": 0.276863496340481, + "instruction_rms": 0.2767469030771587, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 94239712.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.499591748408574e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6659.09619140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.767333984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.53852081298828, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827451.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477800.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1614.7198486328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25662.12890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 450.80853271484375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6659.09619140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.768798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.538660049438477, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827451.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2767469030771587, + "step": 172, + "teaching_rms": 0.276863496340481, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.019123077392578, + "innovation_rms": 0.2567116537650224, + "instruction_rms": 0.2565338532666274, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 84815744.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3496325185921352e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5963.087890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.8902587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.269527435302734, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827738.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477888.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1453.247802734375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25663.025390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 405.7276611328125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5963.087890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.8917236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.269668579101562, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827738.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2565338532666274, + "step": 173, + "teaching_rms": 0.2567116537650224, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.127912998199463, + "innovation_rms": 0.2642271291224182, + "instruction_rms": 0.2641103934156552, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 76334168.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2146692529890263e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5336.6201171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.982177734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.227231979370117, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827908.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 477967.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1307.9229736328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25663.83203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 365.1548767089844, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5336.6201171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.983642578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.227371215820312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827908.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2641103934156552, + "step": 174, + "teaching_rms": 0.2642271291224182, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.1852471828460693, + "innovation_rms": 0.2978679855502292, + "instruction_rms": 0.2978165663744342, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 68700752.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0932023551779144e+18, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4772.74609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.046142578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.38898277282715, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827973.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478038.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1177.130615234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25664.55859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 328.6393737792969, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4772.74609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.047607421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.389123916625977, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827973.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2978165663744342, + "step": 175, + "teaching_rms": 0.2978679855502292, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.0689573287963867, + "innovation_rms": 0.27477864881462377, + "instruction_rms": 0.2747028713300193, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 61830676.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.838821059162276e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4265.21142578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.734395027160645, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827943.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478102.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1059.41748046875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25665.212890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 295.7754211425781, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4265.21142578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.08642578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.734536170959473, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827943.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2747028713300193, + "step": 176, + "teaching_rms": 0.27477864881462377, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.4559617042541504, + "innovation_rms": 0.28780160467903726, + "instruction_rms": 0.2877768807582832, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 55647608.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.854938884526572e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3808.387939453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.1011962890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.245118141174316, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827828.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478160.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 953.4757080078125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25665.80078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 266.1978759765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3808.387939453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.1026611328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.245260238647461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827828.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2877768807582832, + "step": 177, + "teaching_rms": 0.28780160467903726, + "traffic_rms": 0.0 + }, + { + "batch_loss": 3.0356574058532715, + "innovation_rms": 0.279009028345581, + "instruction_rms": 0.2789587593754558, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 50082848.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.969445133512868e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3397.208740234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0970458984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.90463638305664, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827636.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478212.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 858.1281127929688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25666.330078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 239.5780792236328, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3397.208740234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0985107421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.904778480529785, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827636.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27895875937545755, + "step": 178, + "teaching_rms": 0.279009028345581, + "traffic_rms": 3.1193397810880315e-08 + }, + { + "batch_loss": 2.94931960105896, + "innovation_rms": 3206.2685510452966, + "instruction_rms": 3206.2685505553663, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 45082540.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.172500551442104e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3027.113525390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0745849609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.698083877563477, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827375.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478259.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 772.3153076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25666.806640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 215.62026977539062, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3027.113525390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.0760498046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.698226928710938, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827375.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3206.2685510584815, + "step": 179, + "teaching_rms": 3206.2685510452966, + "traffic_rms": 0.05680004858165525 + }, + { + "batch_loss": 2.8555386066436768, + "innovation_rms": 10899.010887625855, + "instruction_rms": 10899.010887361406, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 40581720.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.455250633736847e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55532.4609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.03564453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.612079620361328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827052.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478301.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 695.083740234375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25667.236328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 200.41537475585938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55532.4609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1874.037109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.612222671508789, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8827052.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 10899.010887597187, + "step": 180, + "teaching_rms": 10899.010887625855, + "traffic_rms": 0.07169058495221262 + }, + { + "batch_loss": 53.69892883300781, + "innovation_rms": 1632683.8899226976, + "instruction_rms": 1632683.8899226934, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36528344.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.809725295485256e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49978.671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9818115234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.634580612182617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826673.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478339.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 625.5753784179688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25667.623046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 440.19635009765625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49978.671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9832763671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.634723663330078, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826673.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1632683.8899226945, + "step": 181, + "teaching_rms": 1632683.8899226976, + "traffic_rms": 0.057424240030186625 + }, + { + "batch_loss": 2.85539174079895, + "innovation_rms": 1437.0915126469017, + "instruction_rms": 1437.0915047528385, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32879792.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.2287526284977766e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44979.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9146728515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.754745006561279, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826243.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478373.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 563.017822265625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25667.970703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 754.9651489257812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44979.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.9161376953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.75488805770874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8826243.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1437.0915062697325, + "step": 182, + "teaching_rms": 1437.0915126469017, + "traffic_rms": 0.0660287638007174 + }, + { + "batch_loss": 923.4841918945312, + "innovation_rms": 3788100.555174392, + "instruction_rms": 3788100.5551743917, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 29592286.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.705877468727214e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40480.28125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.83544921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.476913452148438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478403.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 506.7160339355469, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25668.283203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 679.4686279296875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40480.28125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.8369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.476882934570312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 3788100.555174392, + "step": 183, + "teaching_rms": 3788100.555174392, + "traffic_rms": 0.03765189622279445 + }, + { + "batch_loss": 404.11505126953125, + "innovation_rms": 501391.4005809951, + "instruction_rms": 501391.4005809878, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 26635186.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.235289584415539e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36430.34765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.7454833984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.724916458129883, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825252.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478431.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 456.04443359375, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25668.564453125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 611.521728515625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36430.34765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.7469482421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.724885940551758, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8825252.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 501391.40058099496, + "step": 184, + "teaching_rms": 501391.4005809951, + "traffic_rms": 0.08499668843558371 + }, + { + "batch_loss": 355.9176940917969, + "innovation_rms": 891291.8584684258, + "instruction_rms": 891291.8584684238, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 23971398.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8117604541752934e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32785.04296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.645751953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.949853897094727, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824700.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478456.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 410.4399719238281, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25668.818359375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 550.3695678710938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32785.04296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.647216796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.949827194213867, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824700.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 891291.8584684242, + "step": 185, + "teaching_rms": 891291.8584684258, + "traffic_rms": 0.07881819347698286 + }, + { + "batch_loss": 334.0430908203125, + "innovation_rms": 1209558.4590948278, + "instruction_rms": 1209558.4590948175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 21574060.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.430584511836979e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29503.9375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.5372314453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.34711265563965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824115.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478478.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 369.3959655761719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 517.6212158203125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29503.9375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.5386962890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.34708595275879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8824115.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1209558.4590948238, + "step": 186, + "teaching_rms": 1209558.4590948278, + "traffic_rms": 0.06519848096930188 + }, + { + "batch_loss": 224.1370849609375, + "innovation_rms": 1617794.919163045, + "instruction_rms": 1617794.9191630513, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 19416592.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.0875258888545894e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26550.6484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.4208984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.94455623626709, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823500.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478498.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 332.45635986328125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.251953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 558.976318359375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26550.6484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.42236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.944528579711914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8823500.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1617794.919163052, + "step": 187, + "teaching_rms": 1617794.919163045, + "traffic_rms": 0.0463195458421253 + }, + { + "batch_loss": 220.3536834716797, + "innovation_rms": 874731.0354493351, + "instruction_rms": 874731.0354493349, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 17475652.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7787732312496538e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23892.421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.29736328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.669665336608887, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822858.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478516.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 299.2107238769531, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.4375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 540.0653686523438, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23892.421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.298828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.669637680053711, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822858.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 874731.0354493352, + "step": 188, + "teaching_rms": 874731.0354493351, + "traffic_rms": 0.026226356132063647 + }, + { + "batch_loss": 2.8601067066192627, + "innovation_rms": 9043.650425852305, + "instruction_rms": 9043.65042584984, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 15730073.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5008958909448192e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21499.779296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.16748046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.304486274719238, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822192.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478532.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 269.2896423339844, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.603515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 486.4154968261719, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21499.779296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.1689453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.304458618164062, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8822192.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9043.650425859065, + "step": 189, + "teaching_rms": 9043.650425852305, + "traffic_rms": 0.012991948438177 + }, + { + "batch_loss": 1286.6353759765625, + "innovation_rms": 121198932.52707139, + "instruction_rms": 121198932.52707139, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 14162339.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2508063018503373e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19346.185546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.0318603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.075702667236328, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8821505.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478547.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 242.3606719970703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.75390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 437.77392578125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19346.185546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1873.0333251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.075675010681152, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8821505.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 121198932.52707139, + "step": 190, + "teaching_rms": 121198932.52707139, + "traffic_rms": 0.02878346767642578 + }, + { + "batch_loss": 2.9269120693206787, + "innovation_rms": 0.29738323338221917, + "instruction_rms": 0.29734029032163867, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 12746105.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.025725620125696e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17407.7578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.89111328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.969685554504395, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820798.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478560.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 218.12460327148438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25669.888671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 393.99652099609375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17407.7578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.892578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.969658851623535, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820798.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29751377905242204, + "step": 191, + "teaching_rms": 0.29738323338221917, + "traffic_rms": 0.010158763581116153 + }, + { + "batch_loss": 2.598923683166504, + "innovation_rms": 352878.727046624, + "instruction_rms": 352878.72704661876, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 11485154.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8231529722137805e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29846.158203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.7457275390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.974170684814453, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820074.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478572.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 196.3121337890625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.009765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 360.0180969238281, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29846.158203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.7471923828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.974143981933594, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8820074.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 352878.7270466288, + "step": 192, + "teaching_rms": 352878.727046624, + "traffic_rms": 0.08392932259073368 + }, + { + "batch_loss": 932.2616577148438, + "innovation_rms": 112207387.69580767, + "instruction_rms": 112207387.69580767, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 10339830.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.640837623452795e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26861.82421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.5960693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.078118324279785, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819334.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478583.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 234.41970825195312, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.119140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 644.5219116210938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26861.82421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.5975341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.078091621398926, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8819334.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 112207387.69580767, + "step": 193, + "teaching_rms": 112207387.69580767, + "traffic_rms": 0.013491933412467738 + }, + { + "batch_loss": 29.73762321472168, + "innovation_rms": 142702.85970973683, + "instruction_rms": 142702.85970973506, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 9306102.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4767538954672538e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24175.07421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.442626953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.34222984313965, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8818580.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478593.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 191.2609100341797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.216796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 542.1224975585938, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 24175.07421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.444091796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.34255027770996, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8818580.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 142702.85970973465, + "step": 194, + "teaching_rms": 142702.85970973683, + "traffic_rms": 0.007928532531607906 + }, + { + "batch_loss": 147.31788635253906, + "innovation_rms": 217291.06349472623, + "instruction_rms": 217291.0634947254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8376691.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3290784887406592e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21756.423828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.285888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.536293029785156, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817813.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478601.65625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 149.1409454345703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.3046875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 487.9102478027344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21756.423828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.287353515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.536338806152344, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817813.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 217291.0634947265, + "step": 195, + "teaching_rms": 217291.06349472623, + "traffic_rms": 0.022275275921000467 + }, + { + "batch_loss": 98.02535247802734, + "innovation_rms": 87151.23327453088, + "instruction_rms": 87151.23327452905, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7540019.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1961706398665933e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19579.509765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.1260986328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.383174896240234, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817034.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478609.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 128.8003692626953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.384765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 439.11920166015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19579.509765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1872.1275634765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.383087158203125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8817034.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 87151.23327453154, + "step": 196, + "teaching_rms": 87151.23327453088, + "traffic_rms": 0.02093597297506642 + }, + { + "batch_loss": 110.49617767333984, + "innovation_rms": 9651.219023627074, + "instruction_rms": 9651.219023603622, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6787064.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0765535672899994e+17, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17620.150390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.9635009765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.24050521850586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816245.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478616.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.92032623291016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.45703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 395.207275390625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17620.150390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.9649658203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.2403678894043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8816245.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 9651.219023623473, + "step": 197, + "teaching_rms": 9651.219023627074, + "traffic_rms": 0.01986103000747594 + }, + { + "batch_loss": 139.51123046875, + "innovation_rms": 17968.384378315954, + "instruction_rms": 17968.384378292307, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6109267.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.688982363308032e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15856.587890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.7984619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.284793853759766, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8815447.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478622.75, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.51786804199219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.521484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 355.6865234375, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15856.587890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.7999267578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.284656524658203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8815447.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 17968.384378301376, + "step": 198, + "teaching_rms": 17968.384378315954, + "traffic_rms": 0.018495063933818157 + }, + { + "batch_loss": 309.222900390625, + "innovation_rms": 1516852.2388912716, + "instruction_rms": 1516852.2388912714, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5499099.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.720084126977229e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14269.26171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.6312255859375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.9809627532959, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814640.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478628.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.46914672851562, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.580078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 320.11785888671875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14269.26171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.6326904296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.980825424194336, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8814640.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1516852.2388912714, + "step": 199, + "teaching_rms": 1516852.2388912716, + "traffic_rms": 0.01711390573592764 + }, + { + "batch_loss": 93.79814910888672, + "innovation_rms": 44566.17938654317, + "instruction_rms": 44566.17938653945, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4949894.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.84807605787689e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12840.537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.4620361328125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.999473571777344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813826.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478633.5625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.3681869506836, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.6328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 288.1060791015625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12840.537109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.4635009765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.999500274658203, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813826.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 44566.179386541866, + "step": 200, + "teaching_rms": 44566.17938654317, + "traffic_rms": 0.015386626809295773 + }, + { + "batch_loss": 6.241382122039795, + "innovation_rms": 15861.076889584181, + "instruction_rms": 15861.076889578206, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4455546.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.063268280290509e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11554.57421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.198814392089844, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813005.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478638.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.16688537597656, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.6796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 259.29547119140625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11554.57421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.29248046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.198843002319336, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8813005.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 15861.076889584154, + "step": 201, + "teaching_rms": 15861.076889584181, + "traffic_rms": 0.01376293302916301 + }, + { + "batch_loss": 2.875265121459961, + "innovation_rms": 1980.7366439625214, + "instruction_rms": 1980.7366439359062, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4010560.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.356941581110477e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10397.10546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.118408203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.57800579071045, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8812178.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478642.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 125.1427230834961, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.72265625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 233.3659210205078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10397.10546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1871.119873046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.578034400939941, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8812178.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 1980.7366439730527, + "step": 202, + "teaching_rms": 1980.7366439625214, + "traffic_rms": 0.012130779320862147 + }, + { + "batch_loss": 2.7996561527252197, + "innovation_rms": 2143.531357866749, + "instruction_rms": 2143.5313578481628, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3610000.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.721247337100083e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9355.2919921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.9443359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.119132041931152, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811346.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478646.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 130.58822631835938, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.759765625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 210.02932739257812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9355.2919921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.94580078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.119160652160645, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8811346.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 2143.531357873992, + "step": 203, + "teaching_rms": 2143.531357866749, + "traffic_rms": 0.01052291113050126 + }, + { + "batch_loss": 2.723015308380127, + "innovation_rms": 4644.328874188232, + "instruction_rms": 4644.328874183763, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3249423.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.149122474541056e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8417.576171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.7689208984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.806015014648438, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8810509.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478649.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 127.09407043457031, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 189.02638244628906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8417.576171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.7703857421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.80604362487793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8810509.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 4644.328874192497, + "step": 204, + "teaching_rms": 4644.328874188232, + "traffic_rms": 0.009006375942675771 + }, + { + "batch_loss": 2.8032076358795166, + "innovation_rms": 31318.218841325594, + "instruction_rms": 31318.218841325328, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2924843.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.634210012338586e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8415.279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.5924072265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.624091148376465, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8809667.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478652.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 115.77883911132812, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.826171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 170.1237335205078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8415.279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.5938720703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.624119758605957, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8809667.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 31318.21884132626, + "step": 205, + "teaching_rms": 31318.218841325594, + "traffic_rms": 0.0076401103106344456 + }, + { + "batch_loss": 2.8712570667266846, + "innovation_rms": 0.2764500082549549, + "instruction_rms": 0.2764700626176387, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2632359.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.1707890540544e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8454.7177734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.414794921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.560254096984863, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808822.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478655.09375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 104.2009506225586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.853515625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 153.11135864257812, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8454.7177734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.416259765625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.560282707214355, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8808822.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2764700626176387, + "step": 206, + "teaching_rms": 0.2764500082549549, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.8995115756988525, + "innovation_rms": 0.3022319427536631, + "instruction_rms": 0.3022781422600584, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2369123.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.753709933900595e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490.1279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.2362060546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.602704048156738, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807973.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478657.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 93.78085327148438, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.87890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 137.8002166748047, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8490.1279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.2376708984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.60273265838623, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807973.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3022781422600584, + "step": 207, + "teaching_rms": 0.3022319427536631, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.719717502593994, + "innovation_rms": 0.25785018801185927, + "instruction_rms": 0.2578412468404832, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2132211.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3783390264098816e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8521.912109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.0567626953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.740824222564697, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807121.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478659.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 84.40276336669922, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.900390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 124.02019500732422, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8521.912109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1870.0582275390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.740852355957031, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8807121.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2578412468404832, + "step": 208, + "teaching_rms": 0.25785018801185927, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.8080978393554688, + "innovation_rms": 0.29331971850466637, + "instruction_rms": 0.29334658818648046, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1918989.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.040505059344384e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8550.4326171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8765869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.965054512023926, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8806266.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478661.6875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.96248626708984, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.919921875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 111.61817169189453, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8550.4326171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.8780517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.96508264541626, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8806266.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29334658818648046, + "step": 209, + "teaching_rms": 0.29331971850466637, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.8100006580352783, + "innovation_rms": 0.31482099660098894, + "instruction_rms": 0.3148579413911109, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1727090.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.736454446035763e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8576.015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.69580078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.266792297363281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8805408.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478663.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.3662338256836, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.9375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 100.45635223388672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8576.015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.697265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.266820430755615, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8805408.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3148579413911109, + "step": 210, + "teaching_rms": 0.31482099660098894, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.787254571914673, + "innovation_rms": 0.312496978205725, + "instruction_rms": 0.3125383768151723, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1554381.75, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4628089799573504e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8598.955078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.514404296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.638293743133545, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804548.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478665.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 61.52960968017578, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.953125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 90.41071319580078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8598.955078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.515869140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.638321876525879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8804548.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3125383768151723, + "step": 211, + "teaching_rms": 0.312496978205725, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.6707780361175537, + "innovation_rms": 0.272209829233751, + "instruction_rms": 0.2722155449909677, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1398943.625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2165280390119424e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8619.5146484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.3323974609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0725884437561035, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8803686.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478666.5, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 55.37664794921875, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.96875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.36963653564453, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8619.5146484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.3338623046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0726165771484375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8803686.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2722155449909677, + "step": 212, + "teaching_rms": 0.272209829233751, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.786745309829712, + "innovation_rms": 0.30849719063652636, + "instruction_rms": 0.3085369924984441, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1259049.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9948752995352576e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8637.931640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.14990234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.563403129577637, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8802822.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478667.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 49.83898162841797, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.982421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.23267364501953, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8637.931640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1869.1513671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.563431262969971, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8802822.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3085369924984441, + "step": 213, + "teaching_rms": 0.30849719063652636, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.490892171859741, + "innovation_rms": 0.27242623375465413, + "instruction_rms": 0.2723966703730545, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1133144.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7953878125314048e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8654.419921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.9669189453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.105090618133545, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801957.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478668.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 44.85508346557617, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25670.994140625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 65.9094009399414, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8654.419921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.9683837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.105118751525879, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801957.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2723966703730545, + "step": 214, + "teaching_rms": 0.27242623375465413, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5046255588531494, + "innovation_rms": 0.240116748511423, + "instruction_rms": 0.24006809427879872, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1019829.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6158490527531008e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8669.1728515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.7835693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.69256854057312, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801090.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478670.03125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 40.36957550048828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.00390625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 59.318458557128906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8669.1728515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.7850341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.692596435546875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8801090.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24006809427879872, + "step": 215, + "teaching_rms": 0.240116748511423, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.613438606262207, + "innovation_rms": 0.2947589929951547, + "instruction_rms": 0.2947654671844913, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 917846.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.454264115265536e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682.3642578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.599853515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.3212616443634033, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800222.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478670.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 36.332618713378906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.013671875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 53.38661193847656, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682.3642578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.601318359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.321289300918579, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8800222.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2947654671844913, + "step": 216, + "teaching_rms": 0.2947589929951547, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5097427368164062, + "innovation_rms": 0.28831718111188115, + "instruction_rms": 0.28828831242971514, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 826062.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3088377037389824e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.1494140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.4158935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9870522022247314, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8799353.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478671.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 32.69935607910156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.021484375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 48.047950744628906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.1494140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.4173583984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9870798587799072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8799353.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28828831242971514, + "step": 217, + "teaching_rms": 0.28831718111188115, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.620239019393921, + "innovation_rms": 0.3104054756979925, + "instruction_rms": 0.3104137440924104, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 743456.0, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1779539441025024e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.669921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.2315673828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6862337589263916, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8798483.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478672.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 29.429420471191406, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.029296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 43.2431526184082, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.669921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.2330322265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6862616539001465, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8798483.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3104137440924104, + "step": 218, + "teaching_rms": 0.3104054756979925, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.599306344985962, + "innovation_rms": 0.3015312405059806, + "instruction_rms": 0.30154038821056556, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 669110.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0601585604296704e+16, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.05078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.0469970703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4154703617095947, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8797612.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478673.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 26.48647689819336, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.037109375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 38.9188346862793, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.05078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1868.0484619140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4154982566833496, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8797612.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30154038821056556, + "step": 219, + "teaching_rms": 0.3015312405059806, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.47062087059021, + "innovation_rms": 0.25058714309037916, + "instruction_rms": 0.25052363041756415, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 602199.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9541426614370304.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.40625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.8621826171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1717591285705566, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8796740.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478673.90625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 23.837827682495117, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.04296875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 35.02695083618164, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.40625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.8636474609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1717870235443115, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8796740.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25052363041756415, + "step": 220, + "teaching_rms": 0.25058714309037916, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.531863212585449, + "innovation_rms": 0.30875829962950296, + "instruction_rms": 0.30875452584560603, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 541979.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8587284060307456.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.83984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.67724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.952397346496582, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795867.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478674.46875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 21.454044342041016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.048828125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 31.524255752563477, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.83984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.6787109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.952425241470337, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8795867.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30875452584560603, + "step": 221, + "teaching_rms": 0.30875829962950296, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5695884227752686, + "innovation_rms": 0.27945092925617243, + "instruction_rms": 0.27943629875873455, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 487781.4375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7728555815337984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.4423828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.4920654296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8514926433563232, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794993.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478674.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 19.308639526367188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0546875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 28.371829986572266, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.4423828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.4935302734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8514940738677979, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794993.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27943629875873455, + "step": 222, + "teaching_rms": 0.27945092925617243, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4404056072235107, + "innovation_rms": 0.2750626279261716, + "instruction_rms": 0.27502620557083274, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 439003.28125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6955700072742912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.3067626953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8530925512313843, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794119.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478675.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 17.377775192260742, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.05859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 25.53464698791504, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.3082275390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8530939817428589, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8794119.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27502620557083274, + "step": 223, + "teaching_rms": 0.2750626279261716, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.5614099502563477, + "innovation_rms": 0.26358463843060204, + "instruction_rms": 0.26355341463224524, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 395102.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6260130119155712.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.478515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.121337890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8545138835906982, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8793244.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478675.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.639997482299805, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 22.981182098388672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.478515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1867.122802734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8545154333114624, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8793244.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26355341463224524, + "step": 224, + "teaching_rms": 0.26358463843060204, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.579437494277954, + "innovation_rms": 0.3108100442774422, + "instruction_rms": 0.31083092491413566, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 355592.65625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5634117160927232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.0546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.935791015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8557746410369873, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8792369.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478676.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 14.075997352600098, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.06640625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.683063507080078, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.0546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.937255859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855776071548462, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8792369.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.31083092491413566, + "step": 225, + "teaching_rms": 0.3108100442774422, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.408668041229248, + "innovation_rms": 0.2473100742409287, + "instruction_rms": 0.2472415061541871, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 320033.375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5070705498521600.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.0859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.7501220703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568906784057617, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8791493.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478676.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.668396949768066, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0703125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 18.614757537841797, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.0859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.7515869140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568921089172363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8791493.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2472415061541871, + "step": 226, + "teaching_rms": 0.2473100742409287, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4933738708496094, + "innovation_rms": 0.2925992066039426, + "instruction_rms": 0.2925895470351429, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 288030.03125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4563634680233984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.6259765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.5643310546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8578765392303467, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790617.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478676.84375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.401556968688965, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.07421875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 16.753280639648438, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.6259765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.5657958984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8578780889511108, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8790617.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2925895470351429, + "step": 227, + "teaching_rms": 0.2925992066039426, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4220948219299316, + "innovation_rms": 0.29562838644358086, + "instruction_rms": 0.295593358431202, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 259227.03125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4107271319584768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.724609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.37841796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8587453365325928, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8789741.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.261401176452637, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.076171875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 15.07795238494873, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.724609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.3798828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.858746886253357, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8789741.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.295593358431202, + "step": 228, + "teaching_rms": 0.29562838644358086, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.380352020263672, + "innovation_rms": 0.2618839761268197, + "instruction_rms": 0.26182863774098936, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 233304.328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3696544133939200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.42578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.1925048828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8595086336135864, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8788864.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.235260963439941, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.570157051086426, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.42578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.1939697265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8595101833343506, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8788864.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26182863774098936, + "step": 229, + "teaching_rms": 0.2618839761268197, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4585654735565186, + "innovation_rms": 0.29144988141841616, + "instruction_rms": 0.2914303489615634, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 209973.890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3326889720545280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.76953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.0064697265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8601770401000977, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787987.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.59375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.311734199523926, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.080078125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.213141441345215, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.76953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1866.0079345703125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8601784706115723, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787987.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2914303489615634, + "step": 230, + "teaching_rms": 0.29144988141841616, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.4877355098724365, + "innovation_rms": 0.28743268844788356, + "instruction_rms": 0.2874216380348772, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 188976.5, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2994200748490752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.791015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.8204345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860759973526001, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787110.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.78125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.480560779571533, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.08203125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.991827011108398, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.791015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.8218994140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8607614040374756, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8787110.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2874216380348772, + "step": 231, + "teaching_rms": 0.28743268844788356, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.410398006439209, + "innovation_rms": 0.3091903203571299, + "instruction_rms": 0.3091840876191504, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 170078.84375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2694780593111040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.5224609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.63427734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861266016960144, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8786233.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478677.96875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.732504367828369, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.083984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.892643928527832, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.5224609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.6357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8612674474716187, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8786233.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3091840876191504, + "step": 232, + "teaching_rms": 0.3091903203571299, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.364511251449585, + "innovation_rms": 0.28030154143462943, + "instruction_rms": 0.28026665984351146, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 153070.953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2425302533799936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.9931640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.4481201171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861702799797058, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8785356.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.059253692626953, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.0859375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.903379440307617, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.9931640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.4495849609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8617042303085327, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8785356.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28026665984351146, + "step": 233, + "teaching_rms": 0.28030154143462943, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3918378353118896, + "innovation_rms": 0.274034241004796, + "instruction_rms": 0.2739853198889661, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 137763.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2182772307263488.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.2294921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.261962890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862077236175537, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784479.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4533281326293945, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.087890625, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.013041496276855, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.2294921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.263427734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8620786666870117, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8784479.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2739853198889661, + "step": 234, + "teaching_rms": 0.274034241004796, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.341724395751953, + "innovation_rms": 0.2688285220655465, + "instruction_rms": 0.26875931132572245, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 123987.4765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1964495089958912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.25390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.07568359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623956441879272, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8783601.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.907995223999023, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.08984375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.211737155914307, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.25390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1865.0771484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623970746994019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8783601.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26875931132572245, + "step": 235, + "teaching_rms": 0.2688285220655465, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.383183002471924, + "innovation_rms": 0.28960502818337375, + "instruction_rms": 0.289569602682948, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 111588.7265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1768045567541248.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.0888671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.889404296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8626636266708374, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8782723.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.53125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.417195796966553, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.091796875, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.49056339263916, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.0888671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.890869140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862665057182312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8782723.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.289569602682948, + "step": 236, + "teaching_rms": 0.28960502818337375, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3871207237243652, + "innovation_rms": 0.29276195018589, + "instruction_rms": 0.29273233356127637, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 100429.8515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1591240957100032.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.751953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8628861904144287, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8781845.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.975476026535034, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.8415069580078125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.751953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.70458984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8628876209259033, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8781845.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29273233356127637, + "step": 237, + "teaching_rms": 0.29276195018589, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2803874015808105, + "innovation_rms": 0.26279037099768104, + "instruction_rms": 0.26270405280205833, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 90386.8671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1432116847968256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.2607421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.516845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8630677461624146, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780967.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.71875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.577928304672241, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.2573561668396, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.2607421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.518310546875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8630692958831787, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780967.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26270405280205833, + "step": 238, + "teaching_rms": 0.26279037099768104, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.362626314163208, + "innovation_rms": 0.2648427387627951, + "instruction_rms": 0.26476164119025986, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 81348.1796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1288905190014976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.630859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.33056640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8632125854492188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780089.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.8125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.220135450363159, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.7316203117370605, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.630859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.33203125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863214135169983, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8780089.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26476164119025986, + "step": 239, + "teaching_rms": 0.2648427387627951, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3330788612365723, + "innovation_rms": 0.2663034094427114, + "instruction_rms": 0.2662159426901059, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 73213.359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1160014697857024.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.876953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.144287109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8633244037628174, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8779211.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.8981218338012695, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.258458137512207, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.876953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1864.145751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863325834274292, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8779211.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2662159426901059, + "step": 240, + "teaching_rms": 0.2663034094427114, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.332080602645874, + "innovation_rms": 0.2821871271493446, + "instruction_rms": 0.28211482777303043, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 65892.0234375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1044013201227776.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.009765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.9578857421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634064197540283, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778333.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478678.9375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.608309507369995, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8326122760772705, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.009765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.9593505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863407850265503, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778333.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28211482777303043, + "step": 241, + "teaching_rms": 0.2821871271493446, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3469736576080322, + "innovation_rms": 0.2735907047472714, + "instruction_rms": 0.27351144300281705, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 59302.8203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 939611874394112.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.0419921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634614944458008, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777455.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.0, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.3474783897399902, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.4493510723114014, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8778.0419921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.77294921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634629249572754, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777455.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27351144300281705, + "step": 242, + "teaching_rms": 0.2735907047472714, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3143692016601562, + "innovation_rms": 0.2823324576365145, + "instruction_rms": 0.28226663148077835, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 53372.5390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 845650673532928.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.9833984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.5850830078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863492488861084, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776577.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.0625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.1127305030822754, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1044158935546875, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.9833984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.5865478515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634939193725586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776577.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28226663148077835, + "step": 243, + "teaching_rms": 0.2823324576365145, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3266451358795166, + "innovation_rms": 0.25587522192346895, + "instruction_rms": 0.25577527324758226, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 48035.28515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 761085586046976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.8427734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.398681640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8635016679763794, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775699.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.90145742893219, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.7939741611480713, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.8427734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.400146484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8635032176971436, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775699.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25577527324758226, + "step": 244, + "teaching_rms": 0.25587522192346895, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3292765617370605, + "innovation_rms": 0.2878560406695384, + "instruction_rms": 0.2877940532811175, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 43231.7578125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 684977020731392.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.6279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.2122802734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634912967681885, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774821.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.15625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7113116979599, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.5145766735076904, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.6279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.2137451171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634928464889526, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774821.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2877940532811175, + "step": 245, + "teaching_rms": 0.2878560406695384, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.306917190551758, + "innovation_rms": 0.2782939196186503, + "instruction_rms": 0.27821093053156254, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 38908.58203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 616479338790912.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.3466796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.02587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634634017944336, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773943.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.1875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5401804447174072, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.2631189823150635, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.3466796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1863.02734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634649515151978, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773943.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27821093053156254, + "step": 246, + "teaching_rms": 0.2782939196186503, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31752347946167, + "innovation_rms": 0.2871232895858414, + "instruction_rms": 0.287053244609099, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35022.21875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 554831391490048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.005859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.8394775390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863419532775879, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773065.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.21875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3861624002456665, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.036807060241699, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8777.005859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.8409423828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8634212017059326, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773065.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.287053244609099, + "step": 247, + "teaching_rms": 0.2871232895858414, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3311145305633545, + "innovation_rms": 0.27202173926514006, + "instruction_rms": 0.2719526360200078, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35012.1484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 499348265762816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.611328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.6531982421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8633614778518677, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772187.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.25, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2475460767745972, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8331263065338135, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.611328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.6546630859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8633631467819214, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772187.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2719526360200078, + "step": 248, + "teaching_rms": 0.27202173926514006, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3091084957122803, + "innovation_rms": 0.288511336796889, + "instruction_rms": 0.28842972417326124, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35547.5546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 449413432475648.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.1689453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.4669189453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8632906675338745, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771309.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.28125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1227914094924927, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6498136520385742, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8776.1689453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.4683837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8632922172546387, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771309.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28842972417326124, + "step": 249, + "teaching_rms": 0.288511336796889, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3153395652770996, + "innovation_rms": 0.2949414925454183, + "instruction_rms": 0.2948874978839673, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34791.55859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 404472102649856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.6826171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.2806396484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863208293914795, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770431.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.3125, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0105122327804565, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4848322868347168, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.6826171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.2821044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863209843635559, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770431.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2948874978839673, + "step": 250, + "teaching_rms": 0.2949414925454183, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.290590763092041, + "innovation_rms": 0.2716782705503778, + "instruction_rms": 0.2715491157111968, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34940.1796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 364024885673984.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.1572265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.0943603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8631154298782349, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769553.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.34375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9094609618186951, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3363490104675293, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8775.1572265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1862.0958251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863116979598999, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769553.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2715491157111968, + "step": 251, + "teaching_rms": 0.2716782705503778, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307459592819214, + "innovation_rms": 0.2707191889480825, + "instruction_rms": 0.2706055507265049, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35401.9921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 327622387040256.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.5966796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.9080810546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8630132675170898, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768676.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8185148239135742, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2027140855789185, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.5966796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.9095458984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.863014817237854, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768676.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2706055507265049, + "step": 252, + "teaching_rms": 0.2707191889480825, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3141236305236816, + "innovation_rms": 0.2754055739254139, + "instruction_rms": 0.2752728394127358, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34385.63671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 294860141625344.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.00390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.7218017578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8629027605056763, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767799.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.40625, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7366633415222168, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0824426412582397, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8774.00390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.7232666015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8629043102264404, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767799.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2752728394127358, + "step": 253, + "teaching_rms": 0.2754055739254139, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.294116258621216, + "innovation_rms": 0.28171741524698773, + "instruction_rms": 0.2816322051657996, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34097.47265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 265374134173696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773.3837890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.5355224609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8627846240997314, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766922.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6629970073699951, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9741983413696289, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8773.3837890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.5369873046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8627861738204956, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766922.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2816322051657996, + "step": 254, + "teaching_rms": 0.28171741524698773, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2990899085998535, + "innovation_rms": 0.2859890019766697, + "instruction_rms": 0.2858917584020883, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33732.95703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 238836722434048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.7373046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.3492431640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8626596927642822, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766045.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5966972708702087, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8767784833908081, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.7373046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.3507080078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8626611232757568, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766045.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2858917584020883, + "step": 255, + "teaching_rms": 0.2859890019766697, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31665301322937, + "innovation_rms": 0.24368815934624374, + "instruction_rms": 0.24349149437001938, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33315.2265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 214953046835200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.068359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.1629638671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8625285625457764, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765168.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5370275378227234, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7891005873680115, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8772.068359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1861.1644287109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862529993057251, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765168.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24349149437001938, + "step": 256, + "teaching_rms": 0.24368815934624374, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3101770877838135, + "innovation_rms": 0.26744240905182903, + "instruction_rms": 0.26730449878186263, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33096.58984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 193457742151680.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.3779296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.9766845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623919486999512, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764291.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4833247661590576, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7101905345916748, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8771.3779296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.9781494140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8623933792114258, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764291.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26730449878186263, + "step": 257, + "teaching_rms": 0.26744240905182903, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3142447471618652, + "innovation_rms": 0.27951071880576106, + "instruction_rms": 0.27938617729577986, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33474.26953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 174111967936512.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770.6689453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.7904052734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8622503280639648, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763414.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4349922835826874, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6391714811325073, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8770.6689453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.7918701171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.862251877784729, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763414.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27938617729577986, + "step": 258, + "teaching_rms": 0.27951071880576106, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304365634918213, + "innovation_rms": 0.2811811668360158, + "instruction_rms": 0.2810756124280671, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32797.8125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 156700774498304.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.943359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.604248046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8621042966842651, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762537.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3914930522441864, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5752543210983276, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.943359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.605712890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8621058464050293, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762537.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2810756124280671, + "step": 259, + "teaching_rms": 0.2811811668360158, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3088204860687256, + "innovation_rms": 0.2649943542558928, + "instruction_rms": 0.26483423916319115, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33617.73046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 141030703759360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.2021484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.4180908203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8619542121887207, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761660.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35234373807907104, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.517728865146637, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8769.2021484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.4195556640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8619557619094849, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761660.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26483423916319115, + "step": 260, + "teaching_rms": 0.2649943542558928, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3028502464294434, + "innovation_rms": 0.2709728894991329, + "instruction_rms": 0.270831275005025, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34452.7890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 126927633383424.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768.447265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.23193359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8618005514144897, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760783.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3171093463897705, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4659559726715088, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8768.447265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.2333984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861802101135254, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760783.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.270831275005025, + "step": 261, + "teaching_rms": 0.2709728894991329, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3079583644866943, + "innovation_rms": 0.29599078257607714, + "instruction_rms": 0.2959180724950173, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35037.609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 114234872561664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.6806640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.0457763671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861643671989441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759906.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2853983938694, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41936036944389343, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8767.6806640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1860.0472412109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861645221710205, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759906.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2959180724950173, + "step": 262, + "teaching_rms": 0.29599078257607714, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3060388565063477, + "innovation_rms": 0.30885618376348234, + "instruction_rms": 0.30879741324123106, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34748.1875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 102811383627776.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.9033203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.859619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8614838123321533, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759029.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25685855746269226, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37742432951927185, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.9033203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.861083984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8614853620529175, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759029.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30879741324123106, + "step": 263, + "teaching_rms": 0.30885618376348234, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.320901870727539, + "innovation_rms": 0.28049717103404825, + "instruction_rms": 0.2803658118063922, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34289.52734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 92530246942720.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.1162109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.6734619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8613213300704956, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758152.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23117269575595856, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3396818935871124, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8766.1162109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.6749267578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8613228797912598, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758152.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2803658118063922, + "step": 264, + "teaching_rms": 0.28049717103404825, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3094470500946045, + "innovation_rms": 0.28631695112844857, + "instruction_rms": 0.2862184155626006, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34523.41796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83277218054144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.3193359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.4873046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8611564636230469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757275.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20805542171001434, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3057136833667755, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8765.3193359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.48876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.861158013343811, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757275.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2862184155626006, + "step": 265, + "teaching_rms": 0.28631695112844857, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.300950527191162, + "innovation_rms": 0.2750811971027186, + "instruction_rms": 0.2749649319510312, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34988.1015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74949494571008.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764.5146484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.3011474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8609894514083862, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756399.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18724986910820007, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2751423120498657, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8764.5146484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.3026123046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8609910011291504, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756399.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2749649319510312, + "step": 266, + "teaching_rms": 0.2750811971027186, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3068981170654297, + "innovation_rms": 0.28506620633748586, + "instruction_rms": 0.2849602085039063, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34783.875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67454545952768.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763.703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.1151123046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8608206510543823, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755523.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1685248762369156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2476280778646469, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8763.703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1859.1165771484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860822081565857, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755523.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2849602085039063, + "step": 267, + "teaching_rms": 0.28506620633748586, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2924373149871826, + "innovation_rms": 0.29143503177858515, + "instruction_rms": 0.2913721248033816, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33691.6484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 60709090099200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.884765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.9290771484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8606500625610352, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754647.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1516723781824112, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2228652685880661, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.884765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.9305419921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8606514930725098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754647.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2913721248033816, + "step": 268, + "teaching_rms": 0.29143503177858515, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.306351900100708, + "innovation_rms": 0.2937408764765259, + "instruction_rms": 0.29365765498989604, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33461.734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 54638178992128.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.060546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.7430419921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8604779243469238, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753771.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1365051418542862, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2005787342786789, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8762.060546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.7445068359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8604793548583984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753771.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29365765498989604, + "step": 269, + "teaching_rms": 0.2937408764765259, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.29740834236145, + "innovation_rms": 0.28253340441579106, + "instruction_rms": 0.28245497313527296, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32833.671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 49174359834624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761.2314453125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.5570068359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8603044748306274, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752895.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12285462766885757, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18052086234092712, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8761.2314453125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.5584716796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8603057861328125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752895.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28245497313527296, + "step": 270, + "teaching_rms": 0.28253340441579106, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307465076446533, + "innovation_rms": 0.27256050082409095, + "instruction_rms": 0.2724343230838479, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33170.03515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 44256924270592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760.3974609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.3709716796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860129714012146, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752019.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11056916415691376, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1624687761068344, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8760.3974609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.3724365234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.860131025314331, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752019.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2724343230838479, + "step": 271, + "teaching_rms": 0.27256050082409095, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3073205947875977, + "innovation_rms": 0.2716928646440787, + "instruction_rms": 0.27157304000230575, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33558.015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 39831233101824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.5595703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.1849365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8599538803100586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751143.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0995122417807579, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14622189104557037, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8759.5595703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.1864013671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8599551916122437, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751143.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27157304000230575, + "step": 272, + "teaching_rms": 0.2716928646440787, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307616949081421, + "innovation_rms": 0.25873359532577506, + "instruction_rms": 0.25857651879963345, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33064.77734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 35848108113920.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758.7177734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.9989013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8597769737243652, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750267.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08956101536750793, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13159969449043274, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8758.7177734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1858.0003662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8597782850265503, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750267.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25857651879963345, + "step": 273, + "teaching_rms": 0.25873359532577506, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2996084690093994, + "innovation_rms": 0.2654440730962247, + "instruction_rms": 0.2653359208440962, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32840.265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 32263297302528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.8720703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.81298828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8595991134643555, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749391.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0806049108505249, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11843971908092499, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.8720703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.814453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85960054397583, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749391.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2653359208440962, + "step": 274, + "teaching_rms": 0.2654440730962247, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2980923652648926, + "innovation_rms": 0.2820107644277317, + "instruction_rms": 0.28191254387359455, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32745.013671875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 29036967362560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.0234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.6270751953125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8594205379486084, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748515.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07254441827535629, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10659574717283249, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8757.0234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.6285400390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859421968460083, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748515.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28191254387359455, + "step": 275, + "teaching_rms": 0.2820107644277317, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3035073280334473, + "innovation_rms": 0.27097661932059575, + "instruction_rms": 0.2708429964227785, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32360.703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 26133269577728.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.441162109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859241247177124, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747639.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06528997421264648, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09593617171049118, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8756.171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.442626953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.859242558479309, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747639.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2708429964227785, + "step": 276, + "teaching_rms": 0.27097661932059575, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304506540298462, + "innovation_rms": 0.27539353855180604, + "instruction_rms": 0.27526654116134, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32523.55078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 23519943458816.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755.318359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.2552490234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8590612411499023, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746763.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05876097455620766, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08634255081415176, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8755.318359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.2567138671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8590625524520874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746763.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27526654116134, + "step": 277, + "teaching_rms": 0.27539353855180604, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3166027069091797, + "innovation_rms": 0.25346183147305495, + "instruction_rms": 0.25329076363108294, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32863.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 21167949742080.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754.462890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.0693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.858880639076233, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05288487672805786, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07770829647779465, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8754.462890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1857.07080078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.858881950378418, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25329076363108294, + "step": 278, + "teaching_rms": 0.25346183147305495, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.309544086456299, + "innovation_rms": 0.2959230083189856, + "instruction_rms": 0.29582757603559384, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32780.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19051153719296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753.60546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.8834228515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8586995601654053, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745013.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047596387565135956, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06993746757507324, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8753.60546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.8848876953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8587007522583008, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745013.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29582757603559384, + "step": 279, + "teaching_rms": 0.2959230083189856, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.309030294418335, + "innovation_rms": 0.28179095130804693, + "instruction_rms": 0.28168682139794954, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32445.27734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 17146037927936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.74609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.6976318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8585178852081299, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744138.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0428367480635643, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0629437193274498, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8752.74609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.6990966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8585190773010254, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744138.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28168682139794954, + "step": 280, + "teaching_rms": 0.28179095130804693, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3103232383728027, + "innovation_rms": 0.2785562115471031, + "instruction_rms": 0.27843269505945967, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33404.78125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 15431433715712.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.884765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.5118408203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8583358526229858, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743263.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03855307400226593, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0566493459045887, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.884765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.5133056640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8583370447158813, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743263.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27843269505945967, + "step": 281, + "teaching_rms": 0.2785562115471031, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308316230773926, + "innovation_rms": 0.25801942198563865, + "instruction_rms": 0.2578818488050589, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33439.9375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13888290553856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.0224609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.3260498046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8581534624099731, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742388.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0346977673470974, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05098440870642662, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8751.0224609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.3275146484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8581546545028687, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742388.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2578818488050589, + "step": 282, + "teaching_rms": 0.25801942198563865, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308720588684082, + "innovation_rms": 0.268472879389723, + "instruction_rms": 0.2683484300890217, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32784.33984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12499461079040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750.158203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.1402587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8579707145690918, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741513.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.031227989122271538, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04588596522808075, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8750.158203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1856.1417236328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8579719066619873, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741513.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2683484300890217, + "step": 283, + "teaching_rms": 0.268472879389723, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2981810569763184, + "innovation_rms": 0.2986883620037293, + "instruction_rms": 0.29862880536506686, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33639.69140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11249514446848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749.29296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.9544677734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8577876091003418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740638.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.02810519002377987, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.041297368705272675, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8749.29296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.9559326171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8577888011932373, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740638.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.29862880536506686, + "step": 284, + "teaching_rms": 0.2986883620037293, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3011858463287354, + "innovation_rms": 0.261882739170452, + "instruction_rms": 0.26173333505517854, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32432.119140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10124563316736.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748.4267578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.7686767578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8576041460037231, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739763.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.025294670835137367, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03716763108968735, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8748.4267578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.7701416015625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8576054573059082, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739763.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26173333505517854, + "step": 285, + "teaching_rms": 0.261882739170452, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3011229038238525, + "innovation_rms": 0.2628743676613335, + "instruction_rms": 0.26274281835685265, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31947.896484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9112106565632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.560546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.5828857421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8574204444885254, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738888.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.022765202447772026, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03345086798071861, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8747.560546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.5843505859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.857421875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738888.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26274281835685265, + "step": 286, + "teaching_rms": 0.2628743676613335, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2982208728790283, + "innovation_rms": 0.2729924114839236, + "instruction_rms": 0.2728766239255228, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32557.486328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8200896118784.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746.693359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.397216796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.857236623764038, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738013.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.02048868127167225, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03010578081011772, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8746.693359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.398681640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8572380542755127, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738013.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2728766239255228, + "step": 287, + "teaching_rms": 0.2729924114839236, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.30659818649292, + "innovation_rms": 0.25854979436709347, + "instruction_rms": 0.25840654650975947, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33078.0390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7380806402048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745.8251953125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.2115478515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8570526838302612, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737138.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01843981258571148, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.02709520235657692, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8745.8251953125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.2130126953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8570539951324463, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737138.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25840654650975947, + "step": 288, + "teaching_rms": 0.25854979436709347, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3070290088653564, + "innovation_rms": 0.25505321895885164, + "instruction_rms": 0.2549115440726197, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33573.76953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6642725814272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.9560546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.02587890625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568685054779053, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736264.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.016595831140875816, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.024385681375861168, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.9560546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1855.02734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8568698167800903, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736264.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2549115440726197, + "step": 289, + "teaching_rms": 0.25505321895885164, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3022401332855225, + "innovation_rms": 0.2932418180012371, + "instruction_rms": 0.2931564746868434, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34449.5703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5978453442560.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.0869140625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.8402099609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8566842079162598, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735390.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01493624784052372, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.021947111934423447, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8744.0869140625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.8416748046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8566855192184448, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735390.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2931564746868434, + "step": 290, + "teaching_rms": 0.2932418180012371, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301656723022461, + "innovation_rms": 0.2755677790222609, + "instruction_rms": 0.27547077020824756, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34541.47265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5380607836160.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743.216796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.654541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8564997911453247, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734516.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.013442622497677803, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.019752399995923042, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8743.216796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.656005859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8565011024475098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734516.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27547077020824756, + "step": 291, + "teaching_rms": 0.2755677790222609, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.309624671936035, + "innovation_rms": 0.2813513348871347, + "instruction_rms": 0.281244932119809, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34568.4765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4842546790400.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.3466796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.4688720703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8563152551651, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733642.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.012098359875380993, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.017777159810066223, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8742.3466796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.4703369140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8563165664672852, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733642.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.281244932119809, + "step": 292, + "teaching_rms": 0.2813513348871347, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3066153526306152, + "innovation_rms": 0.24845564314083096, + "instruction_rms": 0.24830350362944323, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34921.21484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4358292111360.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741.4755859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.283203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.856130599975586, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732768.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.010888523422181606, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0159994438290596, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8741.4755859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.28466796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8561317920684814, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732768.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24830350362944323, + "step": 293, + "teaching_rms": 0.24845564314083096, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.310338258743286, + "innovation_rms": 0.27764852112734495, + "instruction_rms": 0.2775432385483914, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34968.3125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3922462769152.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740.6044921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.09765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8559458255767822, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731894.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.009799670428037643, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.014399498701095581, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8740.6044921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1854.09912109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8559470176696777, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731894.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2775432385483914, + "step": 294, + "teaching_rms": 0.27764852112734495, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3000028133392334, + "innovation_rms": 0.25550736144475517, + "instruction_rms": 0.25535681511241937, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35077.203125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3530216439808.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739.7333984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.912109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855760931968689, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731020.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00881970301270485, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.012959548272192478, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8739.7333984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.91357421875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8557621240615845, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731020.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25535681511241937, + "step": 295, + "teaching_rms": 0.25550736144475517, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301285743713379, + "innovation_rms": 0.2879855436216822, + "instruction_rms": 0.287916356482223, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36055.50390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3177194717184.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738.861328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.7265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8555760383605957, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730146.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00793773215264082, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.011663593351840973, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8738.861328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.72802734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8555772304534912, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730146.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.287916356482223, + "step": 296, + "teaching_rms": 0.2879855436216822, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3079209327697754, + "innovation_rms": 0.2727608084014196, + "instruction_rms": 0.2726629434830635, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36218.0546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2859475140608.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.9892578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.541015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855391025543213, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729272.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.007143958937376738, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.01049723383039236, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.9892578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.54248046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.855392336845398, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729272.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2726629434830635, + "step": 297, + "teaching_rms": 0.2727608084014196, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3060178756713867, + "innovation_rms": 0.2630135897804151, + "instruction_rms": 0.2628882910721396, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36449.84375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2573527678976.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.1171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.35546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85520601272583, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728398.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00642956281080842, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.009447510354220867, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8737.1171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.35693359375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8552073240280151, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728398.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2628882910721396, + "step": 298, + "teaching_rms": 0.2630135897804151, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2992184162139893, + "innovation_rms": 0.2552271912255807, + "instruction_rms": 0.25508231115077906, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35710.46484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2316174884864.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.2451171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.169921875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8550208806991577, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727524.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005786606576293707, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.008502759039402008, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8736.2451171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1853.17138671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8550221920013428, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727524.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25508231115077906, + "step": 299, + "teaching_rms": 0.2552271912255807, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3002405166625977, + "innovation_rms": 0.28394911046120735, + "instruction_rms": 0.28384731524364426, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35427.8828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2084557422592.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735.373046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.9844970703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8548357486724854, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726650.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005207945592701435, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.007652482949197292, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8735.373046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.9859619140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8548370599746704, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726650.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28384731524364426, + "step": 300, + "teaching_rms": 0.28394911046120735, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.302999258041382, + "innovation_rms": 0.2768150154591984, + "instruction_rms": 0.2767285861456562, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35124.96875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1876101693440.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734.5, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.799072265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.854650616645813, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725777.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004687150940299034, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.006887234281748533, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8734.5, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.800537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.854651927947998, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725777.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2767285861456562, + "step": 301, + "teaching_rms": 0.2768150154591984, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3106067180633545, + "innovation_rms": 0.2900621462457677, + "instruction_rms": 0.2900022836049087, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34610.984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1688491524096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733.6279296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.6136474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8544654846191406, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724904.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004218435846269131, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.006198510527610779, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8733.6279296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.6151123046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8544667959213257, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724904.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2900022836049087, + "step": 302, + "teaching_rms": 0.2900621462457677, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31005859375, + "innovation_rms": 0.26684435432162296, + "instruction_rms": 0.26676355438347044, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34268.37890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1519642345472.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732.755859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.42822265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8542802333831787, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724031.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0037965921219438314, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005578659474849701, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8732.755859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.4296875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8542815446853638, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724031.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26676355438347044, + "step": 303, + "teaching_rms": 0.26684435432162296, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2979652881622314, + "innovation_rms": 0.2497740389351926, + "instruction_rms": 0.2496351814413671, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34192.37109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1367678124032.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.8828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.2427978515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8540949821472168, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723158.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.003416932886466384, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.005020793527364731, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.8828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.2442626953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8540964126586914, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723158.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2496351814413671, + "step": 304, + "teaching_rms": 0.2497740389351926, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301604747772217, + "innovation_rms": 0.2693353395498597, + "instruction_rms": 0.2692259782619113, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33546.4453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1230910259200.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.009765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.057373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8539097309112549, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722285.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0030752394814044237, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.004518714267760515, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8731.009765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1852.058837890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8539111614227295, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722285.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2692259782619113, + "step": 305, + "teaching_rms": 0.2693353395498597, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3041038513183594, + "innovation_rms": 0.29300589607331934, + "instruction_rms": 0.2929474488406543, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33383.55859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1107819233280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730.13671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.8719482421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8537243604660034, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721412.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002767715370282531, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00406684260815382, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8730.13671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.8734130859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8537259101867676, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721412.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2929474488406543, + "step": 306, + "teaching_rms": 0.29300589607331934, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.299781560897827, + "innovation_rms": 0.23126624276634947, + "instruction_rms": 0.23108656068707362, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33092.43359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 997037309952.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.2646484375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.6866455078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8535391092300415, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720539.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0024909437634050846, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0036601582542061806, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8729.2646484375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.6881103515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8535406589508057, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720539.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.23108656068707362, + "step": 307, + "teaching_rms": 0.23126624276634947, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3011538982391357, + "innovation_rms": 0.2572422240556823, + "instruction_rms": 0.25710206788495377, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 31994.44140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 897333592064.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728.3916015625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.5013427734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.85335373878479, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719666.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002241849433630705, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0032941424287855625, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8728.3916015625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.5028076171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8533554077148438, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719666.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25710206788495377, + "step": 308, + "teaching_rms": 0.2572422240556823, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3031558990478516, + "innovation_rms": 0.28833779074579907, + "instruction_rms": 0.288274531523813, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 32023.662109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 807600259072.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727.5185546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.3160400390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8531684875488281, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718793.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0020176644902676344, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00296472804620862, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8727.5185546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.3175048828125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8531700372695923, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718793.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.288274531523813, + "step": 309, + "teaching_rms": 0.28833779074579907, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.307535409927368, + "innovation_rms": 0.2847519063652429, + "instruction_rms": 0.28467618008819495, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33884.44140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 726840246272.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726.6455078125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.1307373046875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8529831171035767, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717920.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0018158979946747422, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002668255241587758, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8726.6455078125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1851.1322021484375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8529847860336304, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717920.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28467618008819495, + "step": 310, + "teaching_rms": 0.2847519063652429, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3039276599884033, + "innovation_rms": 0.2734248341056874, + "instruction_rms": 0.27333294308844575, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33795.5703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 654156234752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725.7734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.9454345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8527978658676147, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717047.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.001634308136999607, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002401429694145918, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8725.7734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.9468994140625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.852799415588379, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717047.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27333294308844575, + "step": 311, + "teaching_rms": 0.2734248341056874, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3061821460723877, + "innovation_rms": 0.2699133393804393, + "instruction_rms": 0.2698119828201293, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 33904.6875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 588740624384.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.900390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.7601318359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8526124954223633, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716175.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0014708773232996464, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.002161286771297455, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.900390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.7615966796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.852614164352417, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716175.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2698119828201293, + "step": 312, + "teaching_rms": 0.2699133393804393, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3083369731903076, + "innovation_rms": 0.270467607500423, + "instruction_rms": 0.27037651337540103, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34267.34375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 529866555392.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.02734375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.5748291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8524272441864014, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715303.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0013237895909696817, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0019451580010354519, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8724.02734375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.5762939453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8524287939071655, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715303.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27037651337540103, + "step": 313, + "teaching_rms": 0.270467607500423, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.322415590286255, + "innovation_rms": 0.3338051136361951, + "instruction_rms": 0.33384327996122615, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34830.48046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 476879912960.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723.154296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.3896484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8522419929504395, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714431.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0011914105853065848, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0017506422009319067, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8723.154296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.39111328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8522435426712036, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714431.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.33384327996122615, + "step": 314, + "teaching_rms": 0.3338051136361951, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2953264713287354, + "innovation_rms": 0.2755120156247644, + "instruction_rms": 0.27542138552778467, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34438.2109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 429191921664.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.2822265625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.2044677734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.852056622505188, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713559.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0010722695151343942, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0015755778877064586, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8722.2822265625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.2059326171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8520581722259521, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713559.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27542138552778467, + "step": 315, + "teaching_rms": 0.2755120156247644, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3050618171691895, + "innovation_rms": 0.3016880662983754, + "instruction_rms": 0.30163777944227105, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35195.21875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 386272722944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721.4091796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.019287109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.851871371269226, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712687.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0009650425636209548, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.001418020110577345, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8721.4091796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1850.020751953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8518729209899902, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712687.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.30163777944227105, + "step": 316, + "teaching_rms": 0.3016880662983754, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.301314353942871, + "innovation_rms": 0.26298608826522385, + "instruction_rms": 0.26286772684863585, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35485.23046875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 347645444096.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720.537109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.8341064453125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8516861200332642, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711815.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0008685382781550288, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0012762180995196104, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8720.537109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.8355712890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8516876697540283, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711815.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26286772684863585, + "step": 317, + "teaching_rms": 0.26298608826522385, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.306528091430664, + "innovation_rms": 0.2684217920832563, + "instruction_rms": 0.26832261503382776, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35866.52734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 312880889856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719.6650390625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.64892578125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8515008687973022, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710943.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007816844154149294, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0011485962895676494, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8719.6650390625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.650390625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8515024185180664, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710943.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26832261503382776, + "step": 318, + "teaching_rms": 0.2684217920832563, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3064236640930176, + "innovation_rms": 0.28972672791674414, + "instruction_rms": 0.2896613653769902, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36255.71484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 281592791040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718.79296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.4637451171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8513156175613403, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710071.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007035159505903721, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0010337366256862879, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8718.79296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.4652099609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8513171672821045, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710071.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2896613653769902, + "step": 319, + "teaching_rms": 0.28972672791674414, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2994956970214844, + "innovation_rms": 0.30577656229897526, + "instruction_rms": 0.3057195223932587, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36013.109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 253433511936.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.9208984375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.2786865234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.851130485534668, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709199.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0006331643671728671, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0009303629631176591, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.9208984375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.2801513671875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8511319160461426, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709199.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.3057195223932587, + "step": 320, + "teaching_rms": 0.30577656229897526, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304568290710449, + "innovation_rms": 0.26431433615154226, + "instruction_rms": 0.26420978220102487, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35666.6328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 228090167296.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.0478515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.0936279296875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.850945234298706, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708327.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0005698479362763464, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0008373266318812966, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8717.0478515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1849.0950927734375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8509467840194702, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708327.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.26420978220102487, + "step": 321, + "teaching_rms": 0.26431433615154226, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.298562526702881, + "innovation_rms": 0.2840294319406367, + "instruction_rms": 0.28394392998665857, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35487.015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 205281148928.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716.17578125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.9085693359375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8507599830627441, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707455.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0005128631019033492, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0007535939221270382, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8716.17578125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.9100341796875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8507615327835083, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707455.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28394392998665857, + "step": 322, + "teaching_rms": 0.2840294319406367, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.304377794265747, + "innovation_rms": 0.25619865800130814, + "instruction_rms": 0.2560767242247034, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36119.25, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 184753029120.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715.3037109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.7235107421875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8505748510360718, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706583.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0004615767684299499, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000678234500810504, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8715.3037109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.7249755859375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8505762815475464, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706583.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2560767242247034, + "step": 323, + "teaching_rms": 0.25619865800130814, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308640718460083, + "innovation_rms": 0.2829524003225242, + "instruction_rms": 0.2828529033989976, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36518.265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 166277726208.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.431640625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.5384521484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8503897190093994, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705712.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00041541908285580575, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0006104110507294536, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8714.431640625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.5399169921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.850391149520874, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705712.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2828529033989976, + "step": 324, + "teaching_rms": 0.2829524003225242, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.29638409614563, + "innovation_rms": 0.2886440562412176, + "instruction_rms": 0.2885618704625799, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36136.7109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 149649948672.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713.5595703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.3533935546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8502044677734375, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704841.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003738771774806082, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0005493699572980404, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8713.5595703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.3548583984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8502060174942017, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704841.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2885618704625799, + "step": 325, + "teaching_rms": 0.2886440562412176, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308162212371826, + "innovation_rms": 0.28761125887870703, + "instruction_rms": 0.2875001420595795, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35795.26171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 134684950528.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712.6875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.1683349609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8500193357467651, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703970.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00033648943644948304, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0004944329266436398, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8712.6875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1848.1697998046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8500207662582397, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703970.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2875001420595795, + "step": 326, + "teaching_rms": 0.28761125887870703, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.319580554962158, + "innovation_rms": 0.2784556520784294, + "instruction_rms": 0.278348337911314, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36360.140625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 121216458752.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711.8154296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.9833984375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8498342037200928, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703099.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003028404898941517, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00044498962233774364, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8711.8154296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.98486328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8498356342315674, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703099.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.278348337911314, + "step": 327, + "teaching_rms": 0.2784556520784294, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.300400495529175, + "innovation_rms": 0.27360456222717955, + "instruction_rms": 0.27348621005629936, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35162.87890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 109094813696.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.943359375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.7984619140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8496490716934204, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702228.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0002725564409047365, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00040049065137282014, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.943359375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.7999267578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.849650502204895, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702228.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27348621005629936, + "step": 328, + "teaching_rms": 0.27360456222717955, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.297814130783081, + "innovation_rms": 0.2632616238292976, + "instruction_rms": 0.2631475324358249, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34813.859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 98185330688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.0712890625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.613525390625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8494640588760376, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701357.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0002453007909934968, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00036044156877323985, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8710.0712890625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.614990234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8494653701782227, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701357.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2631475324358249, + "step": 329, + "teaching_rms": 0.2632616238292976, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3162481784820557, + "innovation_rms": 0.25657891046474496, + "instruction_rms": 0.2564170474954674, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34452.4921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 88366800896.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709.19921875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.4285888671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8492789268493652, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700486.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00022077070025261492, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0003243974060751498, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8709.19921875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.4300537109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8492803573608398, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700486.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2564170474954674, + "step": 330, + "teaching_rms": 0.25657891046474496, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.303004741668701, + "innovation_rms": 0.26801417412158485, + "instruction_rms": 0.2678893762942696, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34831.4609375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79530123264.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708.328125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.24365234375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8490937948226929, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699615.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00019869362586177886, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0002919576654676348, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8708.328125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.2451171875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.849095344543457, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699615.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2678893762942696, + "step": 331, + "teaching_rms": 0.26801417412158485, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3040096759796143, + "innovation_rms": 0.28959446885988255, + "instruction_rms": 0.28952445783121783, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35646.390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71577108480.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707.45703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.0587158203125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8489086627960205, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698744.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0001788242516340688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00026276189601048827, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8707.45703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1847.0601806640625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8489103317260742, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698744.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.28952445783121783, + "step": 332, + "teaching_rms": 0.28959446885988255, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3032758235931396, + "innovation_rms": 0.2788685639031087, + "instruction_rms": 0.2787727130123362, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36826.171875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 64419397632.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706.5849609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.8739013671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8487236499786377, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697873.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00016094182501547039, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00023648570640943944, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8706.5849609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.8753662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8487251996994019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697873.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2787727130123362, + "step": 333, + "teaching_rms": 0.2788685639031087, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.305079698562622, + "innovation_rms": 0.27920223141420836, + "instruction_rms": 0.2790831980520559, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36708.72265625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 57977458688.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705.7138671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.6890869140625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8485386371612549, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697002.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0001448476396035403, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00021283712703734636, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8705.7138671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.6905517578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.848540186882019, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697002.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2790831980520559, + "step": 334, + "teaching_rms": 0.27920223141420836, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3044536113739014, + "innovation_rms": 0.2826387571784897, + "instruction_rms": 0.2825424242620217, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36661.52734375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 52179714048.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.841796875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.5042724609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.848353624343872, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696132.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00013036286691203713, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00019155340851284564, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8704.841796875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.5057373046875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8483551740646362, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696132.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2825424242620217, + "step": 335, + "teaching_rms": 0.2826387571784897, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.319319486618042, + "innovation_rms": 0.2617725870451568, + "instruction_rms": 0.261639136719563, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36241.953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 46961741824.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.970703125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.3194580078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8481686115264893, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695262.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00011732657731045038, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00017239806766156107, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.970703125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.3209228515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8481701612472534, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695262.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.261639136719563, + "step": 336, + "teaching_rms": 0.2617725870451568, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.302999496459961, + "innovation_rms": 0.2714271961521836, + "instruction_rms": 0.27130711089729176, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 37043.05859375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 42265567232.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.099609375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.1346435546875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8479835987091064, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694392.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00010559391375863925, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00015515825361944735, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8703.099609375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1846.1361083984375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8479851484298706, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694392.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27130711089729176, + "step": 337, + "teaching_rms": 0.2714271961521836, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.31793212890625, + "innovation_rms": 0.2484898318317899, + "instruction_rms": 0.2483288116997012, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35756.82421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 38039011328.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702.228515625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.9498291015625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8477987051010132, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693522.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.503452019998804e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00013964241952635348, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8702.228515625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.9512939453125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8478001356124878, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693522.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2483288116997012, + "step": 338, + "teaching_rms": 0.2484898318317899, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3050150871276855, + "innovation_rms": 0.24607951874007386, + "instruction_rms": 0.24589755054209447, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35738.0078125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 34235109376.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701.357421875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.7650146484375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.84761381149292, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692652.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.553106454201043e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.000125678168842569, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8701.357421875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.7664794921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8476152420043945, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692652.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.24589755054209447, + "step": 339, + "teaching_rms": 0.24607951874007386, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3007845878601074, + "innovation_rms": 0.27806233327575625, + "instruction_rms": 0.2779822862528298, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35890.66015625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 30811598848.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700.4873046875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.580322265625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8474289178848267, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691782.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.697795808780938e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00011311034904792905, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8700.4873046875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.581787109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8474303483963013, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691782.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2779822862528298, + "step": 340, + "teaching_rms": 0.27806233327575625, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.314487934112549, + "innovation_rms": 0.22149206750227898, + "instruction_rms": 0.22125733477944715, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35494.07421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 27730438144.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699.6162109375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.3956298828125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8472440242767334, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690912.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.928016227902845e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00010179931268794462, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8699.6162109375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.3970947265625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.847245454788208, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690912.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.22125733477944715, + "step": 341, + "teaching_rms": 0.22149206750227898, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.303394079208374, + "innovation_rms": 0.2772566090447132, + "instruction_rms": 0.2771386569658654, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34833.43359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 24957394944.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698.7451171875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.2109375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8470591306686401, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690042.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.235214095795527e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.161938214674592e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8698.7451171875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.21240234375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8470605611801147, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690042.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2771386569658654, + "step": 342, + "teaching_rms": 0.2772566090447132, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3023366928100586, + "innovation_rms": 0.2804385713347286, + "instruction_rms": 0.2803486557101565, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35683.1328125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 22461655040.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.0262451171875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8468742370605469, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8689172.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.611692540696822e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.245743811130524e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1845.0277099609375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8468756675720215, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8689172.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2803486557101565, + "step": 343, + "teaching_rms": 0.2804385713347286, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.298776865005493, + "innovation_rms": 0.26250410348192543, + "instruction_rms": 0.262372941410382, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36267.2890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20215488512.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.0048828125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.841552734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8466893434524536, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8688302.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.0505230319686234e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.421168993460014e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8697.0048828125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.843017578125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8466907739639282, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8688302.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.262372941410382, + "step": 344, + "teaching_rms": 0.26250410348192543, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.302884340286255, + "innovation_rms": 0.28336259154925825, + "instruction_rms": 0.2832701367846047, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35647.71484375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19763480576.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696.134765625, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.6568603515625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8465044498443604, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8687432.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.5454704377334565e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.679051875835285e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8696.134765625, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.6583251953125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.846505880355835, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8687432.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2832701367846047, + "step": 345, + "teaching_rms": 0.28336259154925825, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2959072589874268, + "innovation_rms": 0.2576541098958782, + "instruction_rms": 0.257547023226699, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35653.3515625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19728865280.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695.263671875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.47216796875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.846319556236267, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8686562.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.0909231756813824e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.01114661549218e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8695.263671875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.4736328125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8463211059570312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8686562.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.257547023226699, + "step": 346, + "teaching_rms": 0.2576541098958782, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3029963970184326, + "innovation_rms": 0.2647267797980056, + "instruction_rms": 0.2645975324370069, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36131.390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19810314240.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.3935546875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.28759765625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8461347818374634, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8685693.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.681830639834516e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4100317356642336e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8694.3935546875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.2890625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8461363315582275, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8685693.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2645975324370069, + "step": 347, + "teaching_rms": 0.2647267797980056, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3057830333709717, + "innovation_rms": 0.25457121986908426, + "instruction_rms": 0.25443099794438656, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35668.6796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19953399808.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693.5234375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.10302734375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8459501266479492, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8684824.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.313647539471276e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.8690282710595056e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8693.5234375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1844.1044921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8459515571594238, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8684824.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.25443099794438656, + "step": 348, + "teaching_rms": 0.25457121986908426, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.3068795204162598, + "innovation_rms": 0.2829021781292106, + "instruction_rms": 0.2828062739936588, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35843.70703125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19933077504.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692.6533203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.91845703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8457653522491455, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683955.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9822827855241485e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.3821251892950386e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8692.6533203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.919921875, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8457667827606201, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683955.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2828062739936588, + "step": 349, + "teaching_rms": 0.2829021781292106, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.303715229034424, + "innovation_rms": 0.2747147069271108, + "instruction_rms": 0.27461643421075116, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 36015.66796875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19901931520.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691.783203125, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.73388671875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8455805778503418, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683086.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6840543796424754e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.943912452086806e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8691.783203125, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.7353515625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8455820083618164, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8683086.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.27461643421075116, + "step": 350, + "teaching_rms": 0.2747147069271108, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.308358907699585, + "innovation_rms": 0.29625551387412313, + "instruction_rms": 0.2961951705813059, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 35441.57421875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19939577856.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.9130859375, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.54931640625, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.845395803451538, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682217.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4156488507287577e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.549520988599397e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.9130859375, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.55078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8453972339630127, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8682217.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2961951705813059, + "step": 351, + "teaching_rms": 0.29625551387412313, + "traffic_rms": 0.0 + }, + { + "batch_loss": 2.2961888313293457, + "innovation_rms": 0.2515914459119246, + "instruction_rms": 0.2514413205977138, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 34457.14453125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 19918106624.0, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.04296875, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.36474609375, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8452110290527344, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681348.0, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 478679.4375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.174083965655882e-05, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 25671.09375, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.1945688533596694e-05, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 8690.04296875, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 1843.3662109375, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.845212459564209, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 8681348.0, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03887232393026352, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12364237010478973, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.2514413205977138, + "step": 352, + "teaching_rms": 0.2515914459119246, + "traffic_rms": 0.0 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} |