diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 17:05:08 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 17:05:08 -0500 |
| commit | 6b85249c718cb366b13355a09e0765cb40c4399d (patch) | |
| tree | 006af80bd1f11c5825daea5b05ad52df909b7aab /results | |
| parent | 8b28522305612d6300f485e14046ae38d49af426 (diff) | |
results: pass dynamic projection training gate
Diffstat (limited to 'results')
| -rw-r--r-- | results/kp_dynamic_projection/dynamic.json | 36859 | ||||
| -rw-r--r-- | results/kp_dynamic_projection_gate.json | 42 |
2 files changed, 36901 insertions, 0 deletions
diff --git a/results/kp_dynamic_projection/dynamic.json b/results/kp_dynamic_projection/dynamic.json new file mode 100644 index 0000000..0a7ec53 --- /dev/null +++ b/results/kp_dynamic_projection/dynamic.json @@ -0,0 +1,36859 @@ +{ + "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": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "max_steps": 352, + "neutral_projection": true, + "predictor_every": 0, + "predictor_mode": "closed_form", + "predictor_warmup": { + "closed_form_fit": { + "max_absolute_residual_soma_slope": 1.6851760875624677e-08, + "max_applied_stability_margin": 0.0, + "max_positive_residual_soma_slope": 1.6851760875624677e-08, + "min_residual_soma_slope": -1.5983413703679616e-08, + "mse": 1.477951640859762e-18, + "observations": 64, + "residual_traffic_rms_ratio": 7.21125499270486e-08, + "stability_margin": 0.0 + }, + "first_mse": 1.477951640859762e-18, + "last_mse": 1.477951640859762e-18, + "post_warmup_traffic_residual_rms_ratio": 7.21125499270486e-08, + "steps": 1, + "task_loader_state_restored": true + }, + "protocol": "kp_dynamic_neutral_projection_diagnosis_v1", + "provenance": { + "git_commit": "8b28522305612d6300f485e14046ae38d49af426", + "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": 0.0, + "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, + 0.0023899395018815994, + 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, + 0.012562948279082775, + 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": 0.004215397890235521, + "instruction_rms": 0.00421539791368719, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 9.022268265823641e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 4.786547602719793e-09, + "min_post_projection_soma_slope": -9.022268265823641e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.0255925100276845e-08, + "pre_projection_traffic_rms_ratio": 0.0021767339184552074 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40285658836364746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.9875131845474243, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05347411707043648, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4058851897716522, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6322960257530212, + "nonfinite_indices": [] + }, + "feedback_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8331910371780396, + "nonfinite_indices": [] + }, + "forward_bn_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.00811514537781477, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08115144819021225, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0037773847579956, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03945024311542511, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05994560942053795, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3794170618057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.632293164730072, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1170727014541626, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01731826254949561, + "step": 1, + "teaching_rms": 0.004215397890235521, + "traffic_rms": 0.01680093257108933 + }, + { + "batch_loss": 2.405984878540039, + "innovation_rms": 0.004090082405056444, + "instruction_rms": 0.004090082400222403, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.0293341763317585, + "max_absolute_post_projection_soma_slope": 1.0328012223226324e-08, + "max_absolute_pre_projection_soma_slope": 0.0293341763317585, + "max_positive_post_projection_soma_slope": 7.367217058629194e-09, + "min_post_projection_soma_slope": -1.0328012223226324e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.0130266765281276e-08, + "pre_projection_traffic_rms_ratio": 0.002087826817196517 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.765994131565094, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.962554454803467, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05457054823637009, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.414070188999176, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3981538414955139, + "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": 0.014180810190737247, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.062230080366134644, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0087769031524658, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.049995459616184235, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08738796412944794, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.378882497549057, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3981483578681946, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1178518533706665, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017318795732276838, + "step": 2, + "teaching_rms": 0.004090082405056444, + "traffic_rms": 0.01683054095208978 + }, + { + "batch_loss": 2.6363866329193115, + "innovation_rms": 0.004281477014817348, + "instruction_rms": 0.004281477010199404, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 7.1881967045328565e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 3.931244219046448e-09, + "min_post_projection_soma_slope": -7.1881967045328565e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.032565852668684e-08, + "pre_projection_traffic_rms_ratio": 0.002129073856902663 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.090734839439392, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.8032267093658447, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06231163442134857, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4310135245323181, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.312594473361969, + "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": 0.017003772780299187, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0537852980196476, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0126572847366333, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058643512427806854, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13530653715133667, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3962412476539612, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31266966462135315, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1187262535095215, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017367830813290103, + "step": 3, + "teaching_rms": 0.004281477014817348, + "traffic_rms": 0.016834415164126856 + }, + { + "batch_loss": 2.8374767303466797, + "innovation_rms": 0.004139290263729305, + "instruction_rms": 0.004139290266117592, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 5.679409831316207e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 5.679409831316207e-09, + "min_post_projection_soma_slope": -2.4107871254841484e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.023932248438777e-08, + "pre_projection_traffic_rms_ratio": 0.00196627310258359 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3719338178634644, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.7656450271606445, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0760226920247078, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4422962963581085, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23460634052753448, + "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": 0.018890362232923508, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07345592975616455, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0175033807754517, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08378931879997253, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1386357545852661, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3977891504764557, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23459231853485107, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1191861629486084, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01732013992437334, + "step": 4, + "teaching_rms": 0.004139290263729305, + "traffic_rms": 0.016818536157679365 + }, + { + "batch_loss": 2.771104335784912, + "innovation_rms": 0.004096562048311282, + "instruction_rms": 0.0040965620543442745, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 4.01043687148217e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 3.945084259271425e-09, + "min_post_projection_soma_slope": -4.01043687148217e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.0154564294847405e-08, + "pre_projection_traffic_rms_ratio": 0.0016485623084769262 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6261553764343262, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.663724899291992, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0815243124961853, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44851982593536377, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40673279762268066, + "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": 0.026366129517555237, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0845746397972107, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.023878812789917, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09993676096200943, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.133967787027359, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38512277603149414, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40671607851982117, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1195976734161377, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01734894673378062, + "step": 5, + "teaching_rms": 0.004096562048311282, + "traffic_rms": 0.0168622872943146 + }, + { + "batch_loss": 2.6946263313293457, + "innovation_rms": 0.004056109808612282, + "instruction_rms": 0.00405610984943646, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 5.5076418981059305e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 5.069858310946529e-09, + "min_post_projection_soma_slope": -5.5076418981059305e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.017632573204681e-08, + "pre_projection_traffic_rms_ratio": 0.00241904844457677 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8491981029510498, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.541136264801025, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08315491676330566, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45889270305633545, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4651939272880554, + "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": 0.03232424706220627, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.087657630443573, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.034621000289917, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10762163996696472, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13301196694374084, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3918379843235016, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4651615023612976, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1198161840438843, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01734358518379774, + "step": 6, + "teaching_rms": 0.004056109808612282, + "traffic_rms": 0.01686502916201303 + }, + { + "batch_loss": 2.553525447845459, + "innovation_rms": 0.004079503407410633, + "instruction_rms": 0.0040795034037757, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 8.602542678204372e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.449757895486073e-09, + "min_post_projection_soma_slope": -8.602542678204372e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.998821786720036e-08, + "pre_projection_traffic_rms_ratio": 0.0023314480673067984 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.127594232559204, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.243213653564453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08289706707000732, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47374504804611206, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6055684089660645, + "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": 0.036853935569524765, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08969195187091827, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0482397079467773, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13618648052215576, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12469950318336487, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39945188164711, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6055078506469727, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1196571588516235, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01735923523184381, + "step": 7, + "teaching_rms": 0.004079503407410633, + "traffic_rms": 0.016866231693702168 + }, + { + "batch_loss": 2.627490282058716, + "innovation_rms": 0.0039021862999269663, + "instruction_rms": 0.0039021862999269663, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 7.563312642844267e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 7.563312642844267e-09, + "min_post_projection_soma_slope": -4.791230079348452e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.996888523973989e-08, + "pre_projection_traffic_rms_ratio": 0.002081751691049506 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.4107956886291504, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.832217216491699, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1155356764793396, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4930708110332489, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5946133732795715, + "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": 0.04038424417376518, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10928460210561752, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.0680220127105713, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19782263040542603, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18074451386928558, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3935914933681488, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.594597339630127, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1195029020309448, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017378706715720203, + "step": 8, + "teaching_rms": 0.0039021862999269663, + "traffic_rms": 0.016928593923059284 + }, + { + "batch_loss": 2.8515405654907227, + "innovation_rms": 0.003699346025198085, + "instruction_rms": 0.003699346023861928, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334159567952156, + "max_absolute_post_projection_soma_slope": 6.407972374944393e-09, + "max_absolute_pre_projection_soma_slope": 0.029334159567952156, + "max_positive_post_projection_soma_slope": 6.407972374944393e-09, + "min_post_projection_soma_slope": -3.882080434891577e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.013874668306742e-08, + "pre_projection_traffic_rms_ratio": 0.0025871031743759723 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.6903271675109863, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.477347373962402, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13230374455451965, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5291388034820557, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6787700653076172, + "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": 0.04955003410577774, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14853888750076294, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.09475839138031, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2673635482788086, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.276836097240448, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3939048945903778, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6788524389266968, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1195645332336426, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017418117653966707, + "step": 9, + "teaching_rms": 0.003699346025198085, + "traffic_rms": 0.01701827986637152 + }, + { + "batch_loss": 2.592252492904663, + "innovation_rms": 0.003457817248391342, + "instruction_rms": 0.003457817249820829, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334159567952156, + "max_absolute_post_projection_soma_slope": 9.929140176723195e-09, + "max_absolute_pre_projection_soma_slope": 0.029334159567952156, + "max_positive_post_projection_soma_slope": 9.929140176723195e-09, + "min_post_projection_soma_slope": -4.819565635472145e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9759904419067227e-08, + "pre_projection_traffic_rms_ratio": 0.002542390393192405 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 2.9536683559417725, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 9.072253227233887, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16985805332660675, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.562380850315094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5692253112792969, + "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": 0.06676124036312103, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17211204767227173, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1240369081497192, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29278564453125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2351815551519394, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3980870842933655, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5693128705024719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1198512315750122, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017475120330856866, + "step": 10, + "teaching_rms": 0.003457817248391342, + "traffic_rms": 0.017119034693430138 + }, + { + "batch_loss": 2.9280824661254883, + "innovation_rms": 0.0036235266416404666, + "instruction_rms": 0.00362352664846104, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 9.094692110522828e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 7.162144211037003e-09, + "min_post_projection_soma_slope": -9.094692110522828e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.981004700541587e-08, + "pre_projection_traffic_rms_ratio": 0.0023352359607872446 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.218597888946533, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 10.213828086853027, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20139651000499725, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.544249951839447, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7155938744544983, + "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": 0.08712887018918991, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2036762833595276, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1560547351837158, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3278103172779083, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47263771295547485, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41011661291122437, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7155453562736511, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.120662808418274, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017588434511621644, + "step": 11, + "teaching_rms": 0.0036235266416404666, + "traffic_rms": 0.01721095165859471 + }, + { + "batch_loss": 2.343245029449463, + "innovation_rms": 0.003182527327244798, + "instruction_rms": 0.0031825273101602734, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 9.401962763888605e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.038156641352543e-09, + "min_post_projection_soma_slope": -9.401962763888605e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9964798689447344e-08, + "pre_projection_traffic_rms_ratio": 0.0023637341525183666 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.465634822845459, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.316248893737793, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2054046243429184, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5334789752960205, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6053800582885742, + "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": 0.11038649827241898, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2501462996006012, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1867213249206543, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.346829891204834, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4739049971103668, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41332152485847473, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6053293347358704, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1211143732070923, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017595498281684174, + "step": 12, + "teaching_rms": 0.003182527327244798, + "traffic_rms": 0.017297144363787816 + }, + { + "batch_loss": 2.6963772773742676, + "innovation_rms": 0.0036149148504966357, + "instruction_rms": 0.0036149148915175684, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 8.25761325984331e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.377933127130973e-09, + "min_post_projection_soma_slope": -8.25761325984331e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.005100677946742e-08, + "pre_projection_traffic_rms_ratio": 0.0024733255925783577 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.6984872817993164, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.760859489440918, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1999506950378418, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5377258658409119, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40972596406936646, + "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": 0.13228461146354675, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2436152547597885, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2102925777435303, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32466447353363037, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45327305793762207, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4142012298107147, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4098726212978363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1216745376586914, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.017711748355507907, + "step": 13, + "teaching_rms": 0.0036149148504966357, + "traffic_rms": 0.017342134831187843 + }, + { + "batch_loss": 2.6845321655273438, + "innovation_rms": 0.003914723782144481, + "instruction_rms": 0.0039147237884577055, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 6.514854433703476e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 3.6374263601146595e-09, + "min_post_projection_soma_slope": -6.514854433703476e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.002797017541619e-08, + "pre_projection_traffic_rms_ratio": 0.002124815032481234 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 3.9275014400482178, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.742070198059082, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19014666974544525, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.553155243396759, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6376553177833557, + "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": 0.1468926966190338, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30149441957473755, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.232784628868103, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40054917335510254, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.45402929186820984, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4061463475227356, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6376075744628906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1223843097686768, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01789763208749299, + "step": 14, + "teaching_rms": 0.003914723782144481, + "traffic_rms": 0.017466640594403966 + }, + { + "batch_loss": 2.311260461807251, + "innovation_rms": 0.0039849377355113446, + "instruction_rms": 0.0039849377355113446, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 5.018017557034682e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 4.525908980923532e-09, + "min_post_projection_soma_slope": -5.018017557034682e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 3.002817082020099e-08, + "pre_projection_traffic_rms_ratio": 0.002634960417192628 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.14396333694458, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.615395545959473, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.34898650646209717, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.58596271276474, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6198999881744385, + "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": 0.1596691757440567, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3146318197250366, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2510299682617188, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2956208884716034, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47197866439819336, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4078208804130554, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6198101043701172, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1238058805465698, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0179607843624926, + "step": 15, + "teaching_rms": 0.0039849377355113446, + "traffic_rms": 0.01750899447324857 + }, + { + "batch_loss": 2.629979372024536, + "innovation_rms": 0.00337989164784317, + "instruction_rms": 0.00337989164491828, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 1.3021160150117339e-08, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.243934537129235e-09, + "min_post_projection_soma_slope": -1.3021160150117339e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.997407714357595e-08, + "pre_projection_traffic_rms_ratio": 0.002454278216184883 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.342349052429199, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 11.80156135559082, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3170488476753235, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6053658127784729, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5947558879852295, + "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": 0.15933124721050262, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38793322443962097, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.252357006072998, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30867651104927063, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.599919855594635, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4416142404079437, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5950180888175964, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1248825788497925, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018011586284771786, + "step": 16, + "teaching_rms": 0.00337989164784317, + "traffic_rms": 0.017696088420068385 + }, + { + "batch_loss": 2.3571066856384277, + "innovation_rms": 0.002947828557088673, + "instruction_rms": 0.002947828587270994, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 5.216218124104444e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 5.216218124104444e-09, + "min_post_projection_soma_slope": -3.9486112157760545e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9710733271662046e-08, + "pre_projection_traffic_rms_ratio": 0.002733979393223697 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.527952671051025, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 12.069674491882324, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2783377468585968, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.607354462146759, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9672995805740356, + "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": 0.18397796154022217, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37498581409454346, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2517787218093872, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4520924687385559, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4852582812309265, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41227251291275024, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9675677418708801, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1260043382644653, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01815415413254352, + "step": 17, + "teaching_rms": 0.002947828557088673, + "traffic_rms": 0.0179072298699025 + }, + { + "batch_loss": 2.917773485183716, + "innovation_rms": 0.0029308363445884325, + "instruction_rms": 0.002930836337842363, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 8.638849635644874e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 3.130812720186782e-09, + "min_post_projection_soma_slope": -8.638849635644874e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.976782298572641e-08, + "pre_projection_traffic_rms_ratio": 0.0029254941894977004 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.682461261749268, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 13.880156517028809, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2739366292953491, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6036310195922852, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5581350326538086, + "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": 0.20834293961524963, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2839869260787964, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.2753486633300781, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4842366874217987, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.527513325214386, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4269394278526306, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5578615069389343, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.127086877822876, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01837501712191462, + "step": 18, + "teaching_rms": 0.0029308363445884325, + "traffic_rms": 0.018155421661458143 + }, + { + "batch_loss": 2.7043209075927734, + "innovation_rms": 0.0025040254036296544, + "instruction_rms": 0.002504025370071927, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 4.836676392727668e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 3.878287468950248e-09, + "min_post_projection_soma_slope": -4.836676392727668e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9386102370919693e-08, + "pre_projection_traffic_rms_ratio": 0.002483489359158014 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.816571235656738, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 20.255985260009766, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33867257833480835, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5979856252670288, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4683253765106201, + "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": 0.23094868659973145, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2260575294494629, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3195348978042603, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4418622851371765, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5465102791786194, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46180590987205505, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.468047022819519, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1279828548431396, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018575361922138614, + "step": 19, + "teaching_rms": 0.0025040254036296544, + "traffic_rms": 0.018420618023077213 + }, + { + "batch_loss": 2.5609121322631836, + "innovation_rms": 0.0020278763007146945, + "instruction_rms": 0.002027876320214527, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 5.553307147465603e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 5.553307147465603e-09, + "min_post_projection_soma_slope": -4.138775988593579e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9281791304909838e-08, + "pre_projection_traffic_rms_ratio": 0.0026415397242112405 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 4.924509525299072, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 27.693004608154297, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3636978566646576, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6001272201538086, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7503175139427185, + "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": 0.2498021274805069, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18853437900543213, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.355666995048523, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40016552805900574, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4897480905056, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49369093775749207, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7500349283218384, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1286063194274902, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01871026630783569, + "step": 20, + "teaching_rms": 0.0020278763007146945, + "traffic_rms": 0.01860585948084852 + }, + { + "batch_loss": 2.8675317764282227, + "innovation_rms": 0.001918787150029089, + "instruction_rms": 0.0019187871551812037, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 7.23513027267586e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 2.0830843716623804e-09, + "min_post_projection_soma_slope": -7.23513027267586e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.924502947006721e-08, + "pre_projection_traffic_rms_ratio": 0.002825394833440622 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.022574424743652, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 33.888973236083984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25866904854774475, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6161615252494812, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.46092134714126587, + "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": 0.25679633021354675, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2102336436510086, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.3815224170684814, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30935031175613403, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4445129632949829, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5169544816017151, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4609529376029968, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1289805173873901, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018771939236156422, + "step": 21, + "teaching_rms": 0.001918787150029089, + "traffic_rms": 0.01868039491903331 + }, + { + "batch_loss": 2.526850461959839, + "innovation_rms": 0.001720930977555772, + "instruction_rms": 0.001720930980428, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 7.227899168071872e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 7.227899168071872e-09, + "min_post_projection_soma_slope": -7.051307981953414e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9274410560332517e-08, + "pre_projection_traffic_rms_ratio": 0.002618425762489259 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.120086193084717, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 39.68775177001953, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2710334360599518, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6350113153457642, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5203153491020203, + "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": 0.2632831931114197, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2038983404636383, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4063328504562378, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.30336275696754456, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38928502798080444, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.510806679725647, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5203473567962646, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.12945556640625, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018773993085359264, + "step": 22, + "teaching_rms": 0.001720930977555772, + "traffic_rms": 0.01870461311967995 + }, + { + "batch_loss": 2.3863327503204346, + "innovation_rms": 0.0015401914344513672, + "instruction_rms": 0.0015401914408699276, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.0293341763317585, + "max_absolute_post_projection_soma_slope": 1.089628121064834e-08, + "max_absolute_pre_projection_soma_slope": 0.0293341763317585, + "max_positive_post_projection_soma_slope": 4.7226915711462425e-09, + "min_post_projection_soma_slope": -1.089628121064834e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8785318220806588e-08, + "pre_projection_traffic_rms_ratio": 0.0021263092910000247 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.163416385650635, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 47.644935607910156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2781839966773987, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6461324691772461, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5315879583358765, + "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": 0.2740176320075989, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16412846744060516, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.43952214717865, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35035908222198486, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35280895233154297, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49870577454566956, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5314187407493591, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1299618482589722, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018822051157221374, + "step": 23, + "teaching_rms": 0.0015401914344513672, + "traffic_rms": 0.01876281084283782 + }, + { + "batch_loss": 2.449244499206543, + "innovation_rms": 0.0015816179165981706, + "instruction_rms": 0.0015816179212860024, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 7.328705198261787e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 7.328705198261787e-09, + "min_post_projection_soma_slope": -5.6048907737249465e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.9176366099783135e-08, + "pre_projection_traffic_rms_ratio": 0.0022879881336165513 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.248114585876465, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 52.303951263427734, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2358570545911789, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6495546698570251, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.565085232257843, + "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": 0.2853512167930603, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15107841789722443, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4586598873138428, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2522547245025635, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32151785492897034, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4959307909011841, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5650885701179504, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1305382251739502, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018915104817077914, + "step": 24, + "teaching_rms": 0.0015816179165981706, + "traffic_rms": 0.018852179598201964 + }, + { + "batch_loss": 2.3830909729003906, + "innovation_rms": 0.001557218008480971, + "instruction_rms": 0.0015572180180035416, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.0293341726064682, + "max_absolute_post_projection_soma_slope": 7.720029060465095e-09, + "max_absolute_pre_projection_soma_slope": 0.0293341726064682, + "max_positive_post_projection_soma_slope": 7.720029060465095e-09, + "min_post_projection_soma_slope": -7.061592643964332e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8960235388231762e-08, + "pre_projection_traffic_rms_ratio": 0.0027099047979641175 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.308901786804199, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 56.85418701171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20682376623153687, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6517421007156372, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.49760809540748596, + "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": 0.2936587929725647, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15412043035030365, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4730075597763062, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19665414094924927, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29181504249572754, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5067051649093628, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4978174865245819, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1308196783065796, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018981961179522843, + "step": 25, + "teaching_rms": 0.001557218008480971, + "traffic_rms": 0.018919632197693846 + }, + { + "batch_loss": 2.144198179244995, + "innovation_rms": 0.0015115907563735711, + "instruction_rms": 0.0015115907465635629, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334157705307007, + "max_absolute_post_projection_soma_slope": 7.611023811193718e-09, + "max_absolute_pre_projection_soma_slope": 0.029334157705307007, + "max_positive_post_projection_soma_slope": 7.611023811193718e-09, + "min_post_projection_soma_slope": -4.5172501295098755e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.885937529927352e-08, + "pre_projection_traffic_rms_ratio": 0.0026159951265860484 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.327198028564453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 62.08979797363281, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1617572009563446, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6542976498603821, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5439119338989258, + "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": 0.30467650294303894, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15170499682426453, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.4935294389724731, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20521904528141022, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2588220536708832, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5290618538856506, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5441229343414307, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1309412717819214, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.018951414035480393, + "step": 26, + "teaching_rms": 0.0015115907563735711, + "traffic_rms": 0.018892191389817143 + }, + { + "batch_loss": 2.4153800010681152, + "innovation_rms": 0.0015061117945122087, + "instruction_rms": 0.0015061117928712594, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 5.499772637307387e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.51110571120239e-09, + "min_post_projection_soma_slope": -5.499772637307387e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.872926958020879e-08, + "pre_projection_traffic_rms_ratio": 0.0028935975618169844 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.365817070007324, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.6650619506836, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11841897666454315, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6592833995819092, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4054579734802246, + "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": 0.313953697681427, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18981797993183136, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5129979848861694, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19468596577644348, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23085296154022217, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5617048740386963, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4054759740829468, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1311874389648438, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01900946987602701, + "step": 27, + "teaching_rms": 0.0015061117945122087, + "traffic_rms": 0.018951980500720503 + }, + { + "batch_loss": 2.359323740005493, + "innovation_rms": 0.0014820449376536213, + "instruction_rms": 0.00148204495266199, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 7.44716954770297e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.866249346870745e-09, + "min_post_projection_soma_slope": -7.44716954770297e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.873317114417236e-08, + "pre_projection_traffic_rms_ratio": 0.0027246826812366066 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.407567501068115, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.82530975341797, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11642671376466751, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6658188700675964, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3590850234031677, + "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": 0.3232567012310028, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19085219502449036, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5366978645324707, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23699910938739777, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2062782198190689, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5963822603225708, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35910314321517944, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1325751543045044, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019115040990177273, + "step": 28, + "teaching_rms": 0.0014820449376536213, + "traffic_rms": 0.019054102290169125 + }, + { + "batch_loss": 2.1401031017303467, + "innovation_rms": 0.0014447177495167658, + "instruction_rms": 0.0014447177478060836, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 6.4812697431193556e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.4812697431193556e-09, + "min_post_projection_soma_slope": -4.2209631345713206e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.874112068401289e-08, + "pre_projection_traffic_rms_ratio": 0.002949424031563487 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.4534430503845215, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.8165512084961, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14939568936824799, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.671166181564331, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4792526066303253, + "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": 0.3342636823654175, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19884562492370605, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5468155145645142, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18119552731513977, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18558773398399353, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6442908644676208, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4790858030319214, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.134307861328125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01917569658636276, + "step": 29, + "teaching_rms": 0.0014447177495167658, + "traffic_rms": 0.0191201148268809 + }, + { + "batch_loss": 2.32609486579895, + "innovation_rms": 0.0015381116666616929, + "instruction_rms": 0.0015381116634480732, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 4.382014751058705e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 2.43779441078118e-09, + "min_post_projection_soma_slope": -4.382014751058705e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8686113579090783e-08, + "pre_projection_traffic_rms_ratio": 0.0031267802564166228 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.485018253326416, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.85914611816406, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17066657543182373, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6829701662063599, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.499703973531723, + "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": 0.3506087064743042, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17127221822738647, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5501805543899536, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16767844557762146, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1715906411409378, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6485351920127869, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4998186230659485, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1360981464385986, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0191512277783927, + "step": 30, + "teaching_rms": 0.0015381116666616929, + "traffic_rms": 0.019096642864813846 + }, + { + "batch_loss": 2.147045850753784, + "innovation_rms": 0.0013485157552765624, + "instruction_rms": 0.0013485157589420043, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 9.076877915958903e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 9.076877915958903e-09, + "min_post_projection_soma_slope": -4.928116581481845e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8473055063079888e-08, + "pre_projection_traffic_rms_ratio": 0.0032792115154199274 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.505574703216553, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.96788024902344, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15952624380588531, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6933261156082153, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.420327752828598, + "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": 0.36606812477111816, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15459416806697845, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5498249530792236, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14138305187225342, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1602126657962799, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6225886344909668, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4204429090023041, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1382051706314087, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01926030384373302, + "step": 31, + "teaching_rms": 0.0013485157552765624, + "traffic_rms": 0.0192201432999611 + }, + { + "batch_loss": 2.243849277496338, + "innovation_rms": 0.0014371800685941526, + "instruction_rms": 0.0014371800651548437, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 6.525788354139195e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.525788354139195e-09, + "min_post_projection_soma_slope": -3.533677350731068e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8431028396227213e-08, + "pre_projection_traffic_rms_ratio": 0.0033888358598405523 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.519827365875244, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.1031494140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12490083277225494, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6840758323669434, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42465531826019287, + "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": 0.3782164752483368, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12148340046405792, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5562918186187744, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2506251335144043, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1408795863389969, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5801061987876892, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4248242974281311, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.140795111656189, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019306424680480865, + "step": 32, + "teaching_rms": 0.0014371800685941526, + "traffic_rms": 0.019255256742904413 + }, + { + "batch_loss": 2.2241790294647217, + "innovation_rms": 0.0014126446276531439, + "instruction_rms": 0.0014126446346512324, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 6.029186039313572e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 5.7976259348890835e-09, + "min_post_projection_soma_slope": -6.029186039313572e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8400397708917095e-08, + "pre_projection_traffic_rms_ratio": 0.0034941470481136966 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.502575874328613, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.0479507446289, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07842442393302917, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6699469685554504, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3576295077800751, + "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": 0.3897109925746918, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13186387717723846, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.5636272430419922, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16948330402374268, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1513044536113739, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5443263053894043, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35779908299446106, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1429550647735596, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019261994571889495, + "step": 33, + "teaching_rms": 0.0014126446276531439, + "traffic_rms": 0.019209093834881192 + }, + { + "batch_loss": 2.174618721008301, + "innovation_rms": 0.0016925661685581665, + "instruction_rms": 0.0016925661700183474, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 5.998483931790588e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 5.998483931790588e-09, + "min_post_projection_soma_slope": -5.078452325335547e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.856993443766235e-08, + "pre_projection_traffic_rms_ratio": 0.003638259636735987 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.520466327667236, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.68096923828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12428317964076996, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6635596752166748, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27445685863494873, + "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": 0.397088885307312, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14094600081443787, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.592240810394287, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39702528715133667, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18021160364151, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5224915146827698, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2741263508796692, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1445497274398804, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01931064966722318, + "step": 34, + "teaching_rms": 0.0016925661685581665, + "traffic_rms": 0.019233220119095736 + }, + { + "batch_loss": 2.059535026550293, + "innovation_rms": 0.0016664646090551012, + "instruction_rms": 0.0016664646060889979, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 4.31233759812244e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.31233759812244e-09, + "min_post_projection_soma_slope": -3.591563269011999e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8549867495506232e-08, + "pre_projection_traffic_rms_ratio": 0.0034942259942952547 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.61225700378418, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.65668487548828, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08333044499158859, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.659454345703125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2820696532726288, + "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": 0.4001646041870117, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15027844905853271, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6197075843811035, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2746674418449402, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15863347053527832, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5210781693458557, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2817382216453552, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1458903551101685, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019358234838186716, + "step": 35, + "teaching_rms": 0.0016664646090551012, + "traffic_rms": 0.019291048427269884 + }, + { + "batch_loss": 2.175593137741089, + "innovation_rms": 0.001492373731634781, + "instruction_rms": 0.001492373736602946, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 4.9618917863369916e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 4.9618917863369916e-09, + "min_post_projection_soma_slope": -3.122276437395044e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8574636978721248e-08, + "pre_projection_traffic_rms_ratio": 0.003490495303189967 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.702652931213379, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.18575286865234, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06358571350574493, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6556105613708496, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2520712614059448, + "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": 0.4036603569984436, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1484304666519165, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6334608793258667, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13753296434879303, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13644079864025116, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5352997183799744, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25218823552131653, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1472491025924683, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019430076458465204, + "step": 36, + "teaching_rms": 0.001492373731634781, + "traffic_rms": 0.019376533335301843 + }, + { + "batch_loss": 2.0879292488098145, + "innovation_rms": 0.001553823726632056, + "instruction_rms": 0.0015538237298131797, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 1.0694585661497058e-08, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 1.0694585661497058e-08, + "min_post_projection_soma_slope": -4.0692791358765135e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.844471520210096e-08, + "pre_projection_traffic_rms_ratio": 0.003632768484021078 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.773155689239502, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.1776123046875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.061476223170757294, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6494380831718445, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3627052307128906, + "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": 0.40932708978652954, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1576300859451294, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6411645412445068, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11699055880308151, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1321553885936737, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5457700490951538, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3628224730491638, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1482449769973755, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019412863378053324, + "step": 37, + "teaching_rms": 0.001553823726632056, + "traffic_rms": 0.019351898824307358 + }, + { + "batch_loss": 2.0935287475585938, + "innovation_rms": 0.0015601162624591927, + "instruction_rms": 0.0015601162482018725, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 4.360434679995251e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 2.1181132403569336e-09, + "min_post_projection_soma_slope": -4.360434679995251e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8463280119707035e-08, + "pre_projection_traffic_rms_ratio": 0.0031659341605569185 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.824796199798584, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.9259033203125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0785025805234909, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6450591087341309, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39759668707847595, + "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": 0.40349236130714417, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1677989661693573, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6541481018066406, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12983542680740356, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11012297123670578, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5386736989021301, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.39771416783332825, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1487712860107422, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01939758221257941, + "step": 38, + "teaching_rms": 0.0015601162624591927, + "traffic_rms": 0.019335974632290302 + }, + { + "batch_loss": 1.9598214626312256, + "innovation_rms": 0.0014409451779766667, + "instruction_rms": 0.0014409451676857007, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 1.023670925803799e-08, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 2.4656541253165187e-09, + "min_post_projection_soma_slope": -1.023670925803799e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8460525077283316e-08, + "pre_projection_traffic_rms_ratio": 0.00347819265184372 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.875295162200928, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.0286636352539, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0738271176815033, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6451500058174133, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27056360244750977, + "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": 0.40020886063575745, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1886492669582367, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6648075580596924, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12011948227882385, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1298592984676361, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5162609815597534, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.27045249938964844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1493520736694336, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019459508044641693, + "step": 39, + "teaching_rms": 0.0014409451779766667, + "traffic_rms": 0.019405634070983065 + }, + { + "batch_loss": 2.0356321334838867, + "innovation_rms": 0.0014526788230394528, + "instruction_rms": 0.0014526787924159222, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 3.985181518117997e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 3.985181518117997e-09, + "min_post_projection_soma_slope": -2.15981210693883e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8387106390338867e-08, + "pre_projection_traffic_rms_ratio": 0.003397257819381332 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.92755126953125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.43846893310547, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06847672909498215, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6440727710723877, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22866126894950867, + "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": 0.3978704810142517, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17061254382133484, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6721285581588745, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10285535454750061, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12588436901569366, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5078697204589844, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22875770926475525, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.149706482887268, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019440465894451153, + "step": 40, + "teaching_rms": 0.0014526788230394528, + "traffic_rms": 0.019386951444585434 + }, + { + "batch_loss": 2.139221668243408, + "innovation_rms": 0.001410493011867198, + "instruction_rms": 0.0014104929943452892, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 6.170733257704342e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.170733257704342e-09, + "min_post_projection_soma_slope": -5.28826049617237e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.846817492489747e-08, + "pre_projection_traffic_rms_ratio": 0.0035031536267370587 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.959159851074219, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.77490997314453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07285113632678986, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6380424499511719, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3022371530532837, + "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": 0.396687775850296, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15743501484394073, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6757162809371948, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10622409731149673, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11051451414823532, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5005580186843872, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3023337125778198, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1500780582427979, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019397995998292773, + "step": 41, + "teaching_rms": 0.001410493011867198, + "traffic_rms": 0.019347140601037476 + }, + { + "batch_loss": 2.179841995239258, + "innovation_rms": 0.0014192709895431525, + "instruction_rms": 0.0014192709895431525, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 7.518718092569543e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.394187401781437e-09, + "min_post_projection_soma_slope": -7.518718092569543e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8227416677682992e-08, + "pre_projection_traffic_rms_ratio": 0.0035975361394975968 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.989053249359131, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.83971405029297, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06855602562427521, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6294699311256409, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36551129817962646, + "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": 0.40316614508628845, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0934072881937027, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.67463219165802, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1194084882736206, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14369100332260132, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5090897083282471, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36560797691345215, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1509382724761963, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01945346862362254, + "step": 42, + "teaching_rms": 0.0014192709895431525, + "traffic_rms": 0.019404446477940752 + }, + { + "batch_loss": 1.9987671375274658, + "innovation_rms": 0.0013707261930993825, + "instruction_rms": 0.001370726194902407, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 4.457650693012738e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.457650693012738e-09, + "min_post_projection_soma_slope": -1.8054675532397368e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.845431657899413e-08, + "pre_projection_traffic_rms_ratio": 0.003482522128035022 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.035983085632324, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.52879333496094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08009535074234009, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6211528182029724, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35709187388420105, + "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": 0.41234254837036133, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1425704061985016, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6805747747421265, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17439813911914825, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23700988292694092, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.531748354434967, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35707297921180725, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1522610187530518, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019371019862438044, + "step": 43, + "teaching_rms": 0.0013707261930993825, + "traffic_rms": 0.01932212327052586 + }, + { + "batch_loss": 2.084608554840088, + "innovation_rms": 0.0013372740152515531, + "instruction_rms": 0.0013372740170996807, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.0293341763317585, + "max_absolute_post_projection_soma_slope": 1.1922786313789402e-08, + "max_absolute_pre_projection_soma_slope": 0.0293341763317585, + "max_positive_post_projection_soma_slope": 2.8750124503318375e-09, + "min_post_projection_soma_slope": -1.1922786313789402e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8432597053028257e-08, + "pre_projection_traffic_rms_ratio": 0.0036584197802693513 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.076443195343018, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.72271728515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08362549543380737, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6241048574447632, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24608293175697327, + "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": 0.4226171672344208, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1702406406402588, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6939371824264526, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15260371565818787, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20941822230815887, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5187015533447266, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24617981910705566, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.153275966644287, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019409091938735523, + "step": 44, + "teaching_rms": 0.0013372740152515531, + "traffic_rms": 0.019363193850319253 + }, + { + "batch_loss": 2.0479283332824707, + "innovation_rms": 0.0012961102566594175, + "instruction_rms": 0.00129611026428671, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 4.628250227511899e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 3.800882719673382e-09, + "min_post_projection_soma_slope": -4.628250227511899e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8259822476574014e-08, + "pre_projection_traffic_rms_ratio": 0.0038617707976010504 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.111560344696045, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.49092864990234, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07160916924476624, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6303220391273499, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4438909888267517, + "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": 0.4350264072418213, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14792446792125702, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.708943247795105, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15006044507026672, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.154499351978302, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5115865468978882, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.44340693950653076, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1541731357574463, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01947921039432225, + "step": 45, + "teaching_rms": 0.0012961102566594175, + "traffic_rms": 0.019438639143521775 + }, + { + "batch_loss": 2.2080702781677246, + "innovation_rms": 0.0013388254777584473, + "instruction_rms": 0.0013388254759124612, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 6.318465750609903e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 3.975000328892975e-09, + "min_post_projection_soma_slope": -6.318465750609903e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8538616473124158e-08, + "pre_projection_traffic_rms_ratio": 0.004100593622433484 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.150835990905762, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.77668762207031, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0688282772898674, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6343538761138916, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3674977421760559, + "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": 0.44022509455680847, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1214863508939743, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.725235939025879, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16292673349380493, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12964555621147156, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5149255394935608, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.36746934056282043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1551414728164673, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019394668320530696, + "step": 46, + "teaching_rms": 0.0013388254777584473, + "traffic_rms": 0.019351320780766408 + }, + { + "batch_loss": 2.0216901302337646, + "innovation_rms": 0.0011615935673586924, + "instruction_rms": 0.0011615935694863323, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 3.4871128207214497e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 3.4871128207214497e-09, + "min_post_projection_soma_slope": -3.23319415684864e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8223259337467724e-08, + "pre_projection_traffic_rms_ratio": 0.00422378236655505 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.151735305786133, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.31105041503906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05803429335355759, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6399099826812744, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2837628126144409, + "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": 0.444673091173172, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10040950030088425, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7395448684692383, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14308977127075195, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09952595084905624, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5153496265411377, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2837960720062256, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1561405658721924, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019408585443157475, + "step": 47, + "teaching_rms": 0.0011615935673586924, + "traffic_rms": 0.01937557741634265 + }, + { + "batch_loss": 1.9896758794784546, + "innovation_rms": 0.0012097358887480497, + "instruction_rms": 0.001209735894876957, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 6.708383626374825e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.174707690651758e-09, + "min_post_projection_soma_slope": -6.708383626374825e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8289629834205e-08, + "pre_projection_traffic_rms_ratio": 0.004514515500273291 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.198065757751465, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.34487915039062, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0532994270324707, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6411167979240417, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20438548922538757, + "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": 0.4483889937400818, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07433921843767166, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7505848407745361, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11039983481168747, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0812116339802742, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5322643518447876, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20421186089515686, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1573182344436646, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019460094717181596, + "step": 48, + "teaching_rms": 0.0012097358887480497, + "traffic_rms": 0.019425479937972175 + }, + { + "batch_loss": 2.1125223636627197, + "innovation_rms": 0.0012939888129752841, + "instruction_rms": 0.0012939888082004111, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 8.820939534359695e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 8.820939534359695e-09, + "min_post_projection_soma_slope": -5.349123810560741e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8432866558040967e-08, + "pre_projection_traffic_rms_ratio": 0.004370058186897737 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.2361273765563965, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.0341567993164, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0528378039598465, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6354435682296753, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2535235285758972, + "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": 0.4527095854282379, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08694367110729218, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7572742700576782, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09292343258857727, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09949301928281784, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5575993061065674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25334981083869934, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1585009098052979, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01941639338571593, + "step": 49, + "teaching_rms": 0.0012939888129752841, + "traffic_rms": 0.019374182809712294 + }, + { + "batch_loss": 2.0311267375946045, + "innovation_rms": 0.0012419718922881814, + "instruction_rms": 0.0012419718942781241, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 1.3888954875085346e-08, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.890306826155211e-09, + "min_post_projection_soma_slope": -1.3888954875085346e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.834426148502058e-08, + "pre_projection_traffic_rms_ratio": 0.004427275367637037 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.268321990966797, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.30520629882812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08733358979225159, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6327759623527527, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32091811299324036, + "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": 0.45541563630104065, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10930114984512329, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.758104920387268, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10973130166530609, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1499006152153015, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5843638181686401, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32093435525894165, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1597901582717896, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019475573158661188, + "step": 50, + "teaching_rms": 0.0012419718922881814, + "traffic_rms": 0.019435797156803696 + }, + { + "batch_loss": 1.9640262126922607, + "innovation_rms": 0.0012249856683065176, + "instruction_rms": 0.0012249856683065176, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 4.056128766194433e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 4.0306566972958535e-09, + "min_post_projection_soma_slope": -4.056128766194433e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8384348676329633e-08, + "pre_projection_traffic_rms_ratio": 0.004611557224264376 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.306838035583496, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.90898132324219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07720139622688293, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6370787620544434, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38265037536621094, + "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": 0.45904451608657837, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13106386363506317, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.756498098373413, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09966962039470673, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15807965397834778, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5961484909057617, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.38275498151779175, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1612406969070435, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019435073185688614, + "step": 51, + "teaching_rms": 0.0012249856683065176, + "traffic_rms": 0.01939434536661964 + }, + { + "batch_loss": 1.9577577114105225, + "innovation_rms": 0.0012745208928817673, + "instruction_rms": 0.001274520884155713, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 6.003383568042864e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 6.003383568042864e-09, + "min_post_projection_soma_slope": -4.824890265098247e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8334908898519712e-08, + "pre_projection_traffic_rms_ratio": 0.004163357152344283 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.3440632820129395, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.53849029541016, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07152056694030762, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6398406028747559, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4059310555458069, + "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": 0.4620364308357239, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1509966105222702, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7519605159759521, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11610157042741776, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16359591484069824, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5931786894798279, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.40603572130203247, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.162630558013916, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01942837788324423, + "step": 52, + "teaching_rms": 0.0012745208928817673, + "traffic_rms": 0.019383842329854788 + }, + { + "batch_loss": 2.203366994857788, + "innovation_rms": 0.001419769246208773, + "instruction_rms": 0.0014197692496902585, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 1.4801491587945748e-08, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 2.6702764444053173e-09, + "min_post_projection_soma_slope": -1.4801491587945748e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8357771430843737e-08, + "pre_projection_traffic_rms_ratio": 0.004633354731944809 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.357435703277588, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.96515655517578, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06951647996902466, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6353361010551453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21504563093185425, + "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": 0.4623344838619232, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1303330510854721, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7491097450256348, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11699721962213516, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12981842458248138, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5824329257011414, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21495959162712097, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1638838052749634, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01939959522244429, + "step": 53, + "teaching_rms": 0.001419769246208773, + "traffic_rms": 0.019351774425086975 + }, + { + "batch_loss": 2.0677602291107178, + "innovation_rms": 0.0013042488146462377, + "instruction_rms": 0.0013042488146462377, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 4.599706837637996e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 4.290495514425174e-09, + "min_post_projection_soma_slope": -4.599706837637996e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.84333765681592e-08, + "pre_projection_traffic_rms_ratio": 0.00458992631018589 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.36190128326416, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.40584564208984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05891773849725723, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6272470355033875, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4779534935951233, + "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": 0.45913243293762207, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16778627038002014, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.748241901397705, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1263902336359024, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1077224388718605, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5667695999145508, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.47786745429039, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1648380756378174, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01938057312656632, + "step": 54, + "teaching_rms": 0.0013042488146462377, + "traffic_rms": 0.01934410054027773 + }, + { + "batch_loss": 1.8439828157424927, + "innovation_rms": 0.0012441355866584984, + "instruction_rms": 0.0012441355876517393, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 5.806701341981579e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.639731265854152e-09, + "min_post_projection_soma_slope": -5.806701341981579e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.842149763367621e-08, + "pre_projection_traffic_rms_ratio": 0.0043814505199566145 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.365230560302734, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.75824737548828, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05559674650430679, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6367945075035095, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5159881711006165, + "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": 0.45796677470207214, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15830978751182556, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7479140758514404, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1150883212685585, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09583641588687897, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5577954053878784, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5158834457397461, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165579080581665, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019366782004624716, + "step": 55, + "teaching_rms": 0.0012441355866584984, + "traffic_rms": 0.019328663353260268 + }, + { + "batch_loss": 2.0862791538238525, + "innovation_rms": 0.001330090971852771, + "instruction_rms": 0.001330090971852771, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 4.563335043172856e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 3.010472982012402e-09, + "min_post_projection_soma_slope": -4.563335043172856e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8581460247013833e-08, + "pre_projection_traffic_rms_ratio": 0.004305118782704581 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.364031791687012, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.98681640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0705457404255867, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.636790931224823, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.41018015146255493, + "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": 0.45518049597740173, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12545320391654968, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7475436925888062, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1279042512178421, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.096439428627491, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5658993124961853, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4100753664970398, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165771484375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019310493191378602, + "step": 56, + "teaching_rms": 0.001330090971852771, + "traffic_rms": 0.019268524730550046 + }, + { + "batch_loss": 2.054839849472046, + "innovation_rms": 0.0012836006570266225, + "instruction_rms": 0.001283600658952029, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 5.112458900669026e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 5.112458900669026e-09, + "min_post_projection_soma_slope": -4.576826473368101e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.845601758664984e-08, + "pre_projection_traffic_rms_ratio": 0.004412890167126069 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.36336088180542, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.48857879638672, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.054565880447626114, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6350729465484619, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22780567407608032, + "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": 0.4571382403373718, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11941047757863998, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.746710181236267, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09670646488666534, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09129246324300766, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5795931816101074, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2277441918849945, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1659702062606812, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01934156395068217, + "step": 57, + "teaching_rms": 0.0012836006570266225, + "traffic_rms": 0.019297495745687603 + }, + { + "batch_loss": 2.1722848415374756, + "innovation_rms": 0.001363378988835842, + "instruction_rms": 0.001363378992461324, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 8.17803247343818e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 7.447706007468469e-09, + "min_post_projection_soma_slope": -8.17803247343818e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8468549568589603e-08, + "pre_projection_traffic_rms_ratio": 0.004319869335927189 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.369202613830566, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.89398956298828, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07231613248586655, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6268892884254456, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23116913437843323, + "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": 0.45787864923477173, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0896349847316742, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7480559349060059, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08297040313482285, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07229604572057724, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5712304711341858, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23110763728618622, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1661096811294556, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019331772864787326, + "step": 58, + "teaching_rms": 0.001363378988835842, + "traffic_rms": 0.01928439777376985 + }, + { + "batch_loss": 2.0322155952453613, + "innovation_rms": 0.0012261937331915905, + "instruction_rms": 0.0012261937392382362, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 5.609815278972974e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 3.6610701137362867e-09, + "min_post_projection_soma_slope": -5.609815278972974e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.855626345486651e-08, + "pre_projection_traffic_rms_ratio": 0.004182396353433535 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.369305610656738, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.26412963867188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05810612440109253, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6181322932243347, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37132328748703003, + "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": 0.4623059630393982, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07641896605491638, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7478002309799194, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1007504016160965, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0805024802684784, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5579445362091064, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3714883625507355, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1666419506072998, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019290123922337678, + "step": 59, + "teaching_rms": 0.0012261937331915905, + "traffic_rms": 0.01924929332447877 + }, + { + "batch_loss": 1.8605231046676636, + "innovation_rms": 0.0012027443979171122, + "instruction_rms": 0.0012027444020268018, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334159567952156, + "max_absolute_post_projection_soma_slope": 6.162136134690854e-09, + "max_absolute_pre_projection_soma_slope": 0.029334159567952156, + "max_positive_post_projection_soma_slope": 6.162136134690854e-09, + "min_post_projection_soma_slope": -4.487512139661476e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.848724398600853e-08, + "pre_projection_traffic_rms_ratio": 0.004427236654265319 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.37770938873291, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.66402435302734, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.057341091334819794, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6121822595596313, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3775874376296997, + "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": 0.4685760736465454, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06774473190307617, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7470134496688843, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10064635425806046, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07785896211862564, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5526950359344482, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37775254249572754, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.167160987854004, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019306073208706725, + "step": 60, + "teaching_rms": 0.0012027443979171122, + "traffic_rms": 0.019265276279162382 + }, + { + "batch_loss": 2.1853272914886475, + "innovation_rms": 0.0014082057625382182, + "instruction_rms": 0.0014082057467428865, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334168881177902, + "max_absolute_post_projection_soma_slope": 7.646550059803303e-09, + "max_absolute_pre_projection_soma_slope": 0.029334168881177902, + "max_positive_post_projection_soma_slope": 3.024239969562359e-09, + "min_post_projection_soma_slope": -7.646550059803303e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.849852330462549e-08, + "pre_projection_traffic_rms_ratio": 0.004636192474746684 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.349400520324707, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.81746673583984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05863098055124283, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6109216213226318, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16381549835205078, + "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": 0.47201091051101685, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06393401324748993, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.752095103263855, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10917596518993378, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07962452620267868, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5597610473632812, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16412478685379028, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1674938201904297, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019302109958169212, + "step": 61, + "teaching_rms": 0.0014082057625382182, + "traffic_rms": 0.019254461112757137 + }, + { + "batch_loss": 1.9451091289520264, + "innovation_rms": 0.001226442600971853, + "instruction_rms": 0.001226442595934004, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 9.227312247617192e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 9.227312247617192e-09, + "min_post_projection_soma_slope": -2.0556150115425e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.839381864334573e-08, + "pre_projection_traffic_rms_ratio": 0.004204887117367382 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.341182231903076, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.75130462646484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07468307018280029, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6095262169837952, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20436164736747742, + "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": 0.4729664921760559, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09452460706233978, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.760757565498352, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09382472932338715, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07452648878097534, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5689052939414978, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20413021743297577, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.167810082435608, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019276518907398312, + "step": 62, + "teaching_rms": 0.001226442600971853, + "traffic_rms": 0.019240061419906203 + }, + { + "batch_loss": 2.0903494358062744, + "innovation_rms": 0.0014429624606820818, + "instruction_rms": 0.0014429624589693184, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 5.513119738509431e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 2.073142990610677e-09, + "min_post_projection_soma_slope": -5.513119738509431e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.849266715062875e-08, + "pre_projection_traffic_rms_ratio": 0.004644981066734745 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.331137657165527, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.50147247314453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06771472096443176, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6072320938110352, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31783008575439453, + "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": 0.47552356123924255, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14096352458000183, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.765540599822998, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12436577677726746, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07226605713367462, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5792279839515686, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31804144382476807, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1682814359664917, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019261162481588927, + "step": 63, + "teaching_rms": 0.0014429624606820818, + "traffic_rms": 0.01921174640010551 + }, + { + "batch_loss": 1.9356247186660767, + "innovation_rms": 0.001215261280361868, + "instruction_rms": 0.0012152612783281878, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334159567952156, + "max_absolute_post_projection_soma_slope": 5.16232301350783e-09, + "max_absolute_pre_projection_soma_slope": 0.029334159567952156, + "max_positive_post_projection_soma_slope": 5.16232301350783e-09, + "min_post_projection_soma_slope": -2.8467241897089934e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8442463698646486e-08, + "pre_projection_traffic_rms_ratio": 0.004375821844479557 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.302613258361816, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.15662384033203, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06374655663967133, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6013418436050415, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3158697187900543, + "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": 0.4799360930919647, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13182689249515533, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.766046166419983, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14939959347248077, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06443918496370316, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5787648558616638, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31608110666275024, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1688297986984253, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019272383973338726, + "step": 64, + "teaching_rms": 0.001215261280361868, + "traffic_rms": 0.019236045755480703 + }, + { + "batch_loss": 1.9776138067245483, + "innovation_rms": 0.0013010449442470004, + "instruction_rms": 0.0013010449518453635, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 6.4836482849273125e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 6.4836482849273125e-09, + "min_post_projection_soma_slope": -5.269855662959344e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8419542316186648e-08, + "pre_projection_traffic_rms_ratio": 0.004264418437813208 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.285537242889404, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.73570251464844, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08349072188138962, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6017215847969055, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21460406482219696, + "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": 0.4843713641166687, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07811032235622406, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7604652643203735, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16030649840831757, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08336779475212097, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5776510834693909, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21481546759605408, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1695835590362549, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01928470075471668, + "step": 65, + "teaching_rms": 0.0013010449442470004, + "traffic_rms": 0.019240014950516353 + }, + { + "batch_loss": 2.0706210136413574, + "innovation_rms": 0.0012631793458089365, + "instruction_rms": 0.0012631793516785378, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334168881177902, + "max_absolute_post_projection_soma_slope": 7.388119005469207e-09, + "max_absolute_pre_projection_soma_slope": 0.029334168881177902, + "max_positive_post_projection_soma_slope": 4.247244778099457e-09, + "min_post_projection_soma_slope": -7.388119005469207e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8395816241390952e-08, + "pre_projection_traffic_rms_ratio": 0.004476922148233346 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.2561774253845215, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.4673080444336, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1114334836602211, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6050979495048523, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18269005417823792, + "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": 0.4923069477081299, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0793558806180954, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7496203184127808, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20812547206878662, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11134330928325653, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5747249126434326, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1826782077550888, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.17031991481781, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019255734490259708, + "step": 66, + "teaching_rms": 0.0012631793458089365, + "traffic_rms": 0.01921466319757021 + }, + { + "batch_loss": 2.0646421909332275, + "innovation_rms": 0.0014660757293814904, + "instruction_rms": 0.0014660757352816531, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 6.3173413167305625e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.596056868422238e-09, + "min_post_projection_soma_slope": -6.3173413167305625e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8642173460594224e-08, + "pre_projection_traffic_rms_ratio": 0.0042694710205923555 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.238163471221924, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.25798797607422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15754160284996033, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6063733100891113, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22638800740242004, + "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": 0.5017626881599426, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09455733001232147, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7331472635269165, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2023235410451889, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15745143592357635, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5759375095367432, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22687488794326782, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1707597970962524, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01923308976291889, + "step": 67, + "teaching_rms": 0.0014660757293814904, + "traffic_rms": 0.019177881811461784 + }, + { + "batch_loss": 2.023813486099243, + "innovation_rms": 0.0014606991871150517, + "instruction_rms": 0.0014606991904989833, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334168881177902, + "max_absolute_post_projection_soma_slope": 5.815499193317919e-09, + "max_absolute_pre_projection_soma_slope": 0.029334168881177902, + "max_positive_post_projection_soma_slope": 4.842084511125222e-09, + "min_post_projection_soma_slope": -5.815499193317919e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8660785440274265e-08, + "pre_projection_traffic_rms_ratio": 0.004557746446284917 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.214271545410156, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.86249542236328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13964124023914337, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6074854135513306, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3116621971130371, + "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": 0.5112302303314209, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09467564523220062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7455312013626099, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18835869431495667, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1395510733127594, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.580647349357605, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3121490776538849, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.17093026638031, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019169112238212944, + "step": 68, + "teaching_rms": 0.0014606991871150517, + "traffic_rms": 0.019112666366240816 + }, + { + "batch_loss": 2.0444717407226562, + "innovation_rms": 0.0014792327262527025, + "instruction_rms": 0.0014792327262527025, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 7.947975610989033e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 7.947975610989033e-09, + "min_post_projection_soma_slope": -3.2529423599214624e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8716335178805173e-08, + "pre_projection_traffic_rms_ratio": 0.004398283590696287 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.182476043701172, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.57559967041016, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1209404394030571, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6091548204421997, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31524479389190674, + "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": 0.5168942213058472, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06781236827373505, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7598763704299927, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18800543248653412, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12085027247667313, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5783556699752808, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.31573164463043213, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1711851358413696, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019168395295417043, + "step": 69, + "teaching_rms": 0.0014792327262527025, + "traffic_rms": 0.01911271722315763 + }, + { + "batch_loss": 1.905808448791504, + "innovation_rms": 0.0013964508888557065, + "instruction_rms": 0.0013964508897406117, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 3.869548237389608e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 3.869548237389608e-09, + "min_post_projection_soma_slope": -2.4568429513038836e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8654553417887434e-08, + "pre_projection_traffic_rms_ratio": 0.004548597496325731 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.1580023765563965, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.10365295410156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09996835142374039, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6113592982292175, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33913177251815796, + "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": 0.5199089050292969, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08530206978321075, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7685424089431763, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16564372181892395, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09987818449735641, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5859822630882263, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.33961859345436096, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.171368956565857, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019162265068977967, + "step": 70, + "teaching_rms": 0.0013964508888557065, + "traffic_rms": 0.019107568309655883 + }, + { + "batch_loss": 1.9320321083068848, + "innovation_rms": 0.001465022275514693, + "instruction_rms": 0.0014650222788886388, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 5.1616053653447125e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.020763277878814e-09, + "min_post_projection_soma_slope": -5.1616053653447125e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.851211284967033e-08, + "pre_projection_traffic_rms_ratio": 0.004394778894414735 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.131688594818115, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.0160140991211, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06848417967557907, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.614554762840271, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1805766373872757, + "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": 0.5214954018592834, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0701492503285408, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.772570013999939, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13865427672863007, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07700624316930771, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.601292073726654, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18061372637748718, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1714872121810913, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01918165448908542, + "step": 71, + "teaching_rms": 0.001465022275514693, + "traffic_rms": 0.01912501168231255 + }, + { + "batch_loss": 2.038365602493286, + "innovation_rms": 0.0014929338434380576, + "instruction_rms": 0.0014929338484043586, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 4.240503947983143e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 3.4384299851808464e-09, + "min_post_projection_soma_slope": -4.240503947983143e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8680639551354367e-08, + "pre_projection_traffic_rms_ratio": 0.004399022546853951 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.0945916175842285, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.85936737060547, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058398887515068054, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6213354468345642, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28562110662460327, + "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": 0.5198708176612854, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06593038141727448, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7776588201522827, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10335054993629456, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058239322155714035, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6238603591918945, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28538966178894043, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1720043420791626, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019203812041032987, + "step": 72, + "teaching_rms": 0.0014929338434380576, + "traffic_rms": 0.01914678045357038 + }, + { + "batch_loss": 1.9411442279815674, + "innovation_rms": 0.0014543894738120243, + "instruction_rms": 0.0014543894755113303, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 8.006136198446256e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.020015431649426e-09, + "min_post_projection_soma_slope": -8.006136198446256e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8490224822641225e-08, + "pre_projection_traffic_rms_ratio": 0.004460891211160153 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.0543646812438965, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.96240997314453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05508439615368843, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6300584077835083, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2703108787536621, + "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": 0.5222725868225098, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09001652896404266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7755869626998901, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09520624577999115, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08856555819511414, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6347799897193909, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2702059745788574, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1722646951675415, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019268918975383603, + "step": 73, + "teaching_rms": 0.0014543894738120243, + "traffic_rms": 0.019214350870955698 + }, + { + "batch_loss": 1.9747188091278076, + "innovation_rms": 0.0015970872326353917, + "instruction_rms": 0.0015970872372778175, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02933417074382305, + "max_absolute_post_projection_soma_slope": 3.734946130151684e-09, + "max_absolute_pre_projection_soma_slope": 0.02933417074382305, + "max_positive_post_projection_soma_slope": 3.537614645665599e-09, + "min_post_projection_soma_slope": -3.734946130151684e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8515820871014125e-08, + "pre_projection_traffic_rms_ratio": 0.004363186011032362 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.025553226470947, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.79243469238281, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.051590800285339355, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6418644785881042, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22843226790428162, + "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": 0.5208723545074463, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11393319070339203, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.774339199066162, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11061519384384155, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0909322202205658, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6334571838378906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22820086777210236, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1722335815429688, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019209489831570475, + "step": 74, + "teaching_rms": 0.0015970872326353917, + "traffic_rms": 0.019146824119498397 + }, + { + "batch_loss": 1.9740276336669922, + "innovation_rms": 0.0016375058114546817, + "instruction_rms": 0.0016375058144732393, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 9.895084751576633e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 9.895084751576633e-09, + "min_post_projection_soma_slope": -4.524709495967727e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8463342577129374e-08, + "pre_projection_traffic_rms_ratio": 0.0046791647723052835 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.994736194610596, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.37659454345703, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.038404639810323715, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6420072913169861, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21060264110565186, + "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": 0.5174012184143066, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11521422117948532, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7762665748596191, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09466378390789032, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07714926451444626, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6330339908599854, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21040064096450806, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.172551155090332, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01923345150463344, + "step": 75, + "teaching_rms": 0.0016375058114546817, + "traffic_rms": 0.019161866540361262 + }, + { + "batch_loss": 1.9679899215698242, + "innovation_rms": 0.0014226633658439777, + "instruction_rms": 0.0014226633675811793, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.0293341763317585, + "max_absolute_post_projection_soma_slope": 1.3487400529754723e-08, + "max_absolute_pre_projection_soma_slope": 0.0293341763317585, + "max_positive_post_projection_soma_slope": 6.437752109178518e-09, + "min_post_projection_soma_slope": -1.3487400529754723e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8556751819890778e-08, + "pre_projection_traffic_rms_ratio": 0.0043994963718310175 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 5.991239547729492, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.03634643554688, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05352964624762535, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.650120735168457, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18719124794006348, + "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": 0.513435423374176, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12754084169864655, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.770787239074707, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1045481264591217, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09792971611022949, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6386942267417908, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18726514279842377, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1723486185073853, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01926332905112112, + "step": 76, + "teaching_rms": 0.0014226633658439777, + "traffic_rms": 0.019215330474296688 + }, + { + "batch_loss": 1.9574987888336182, + "innovation_rms": 0.001432167963480611, + "instruction_rms": 0.0014321679652062836, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334165155887604, + "max_absolute_post_projection_soma_slope": 7.0205543600820874e-09, + "max_absolute_pre_projection_soma_slope": 0.029334165155887604, + "max_positive_post_projection_soma_slope": 2.970482526620799e-09, + "min_post_projection_soma_slope": -7.0205543600820874e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.850211388719983e-08, + "pre_projection_traffic_rms_ratio": 0.0042955551211078065 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.011597633361816, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.7960433959961, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.057387568056583405, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6588596701622009, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16729095578193665, + "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": 0.5084859728813171, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11638492345809937, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.763824224472046, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08016063272953033, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09916436672210693, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.643409252166748, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16732625663280487, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1722447872161865, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019192907915787767, + "step": 77, + "teaching_rms": 0.001432167963480611, + "traffic_rms": 0.019141083615424073 + }, + { + "batch_loss": 1.8805102109909058, + "innovation_rms": 0.001429343389364886, + "instruction_rms": 0.0014293433859067207, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 6.782981287756229e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 6.566012178410574e-09, + "min_post_projection_soma_slope": -6.782981287756229e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8635006806425632e-08, + "pre_projection_traffic_rms_ratio": 0.004797677757483476 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.027105331420898, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.37762451171875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05712811276316643, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6635767817497253, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21275624632835388, + "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": 0.5057000517845154, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0790550634264946, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7630579471588135, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07675591856241226, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.087901271879673, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6383852958679199, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2127915471792221, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1726799011230469, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019223672999140948, + "step": 78, + "teaching_rms": 0.001429343389364886, + "traffic_rms": 0.01916676308562054 + }, + { + "batch_loss": 1.7949274778366089, + "innovation_rms": 0.001397550664481771, + "instruction_rms": 0.0013975506503344313, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 5.881399811613619e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 5.881399811613619e-09, + "min_post_projection_soma_slope": -3.883813270988412e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8534158828742528e-08, + "pre_projection_traffic_rms_ratio": 0.0045922656985908465 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.041244029998779, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.96879577636719, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05972101911902428, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6663053035736084, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17320320010185242, + "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": 0.5054387450218201, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08169413357973099, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7668482065200806, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0940704196691513, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11540071666240692, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6344838738441467, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17323850095272064, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1727895736694336, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0191968999909303, + "step": 79, + "teaching_rms": 0.001397550664481771, + "traffic_rms": 0.019141577181671525 + }, + { + "batch_loss": 2.008207082748413, + "innovation_rms": 0.0015472003382031004, + "instruction_rms": 0.001547200328618874, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02933417074382305, + "max_absolute_post_projection_soma_slope": 6.2691434266071155e-09, + "max_absolute_pre_projection_soma_slope": 0.02933417074382305, + "max_positive_post_projection_soma_slope": 5.396156410597541e-09, + "min_post_projection_soma_slope": -6.2691434266071155e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.85731524415177e-08, + "pre_projection_traffic_rms_ratio": 0.004250927876045901 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.059412479400635, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.54737854003906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058746203780174255, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6696330308914185, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18410232663154602, + "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": 0.5026572346687317, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0920553132891655, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.77009916305542, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11129918694496155, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06828689575195312, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.639125406742096, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18410325050354004, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.173426866531372, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019182138314118087, + "step": 80, + "teaching_rms": 0.0015472003382031004, + "traffic_rms": 0.019119324199537942 + }, + { + "batch_loss": 1.8842084407806396, + "innovation_rms": 0.0015305118466646002, + "instruction_rms": 0.0015305118466646002, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 4.146721632736217e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 2.3589634690068806e-09, + "min_post_projection_soma_slope": -4.146721632736217e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8391347693272258e-08, + "pre_projection_traffic_rms_ratio": 0.004572318483826007 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.073643207550049, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.23894500732422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0730767548084259, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6730332374572754, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1975351870059967, + "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": 0.5018354058265686, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1122574657201767, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7723133563995361, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09907431155443192, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09097692370414734, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6464501619338989, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1974971443414688, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1742663383483887, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019250057776133637, + "step": 81, + "teaching_rms": 0.0015305118466646002, + "traffic_rms": 0.019189929659833616 + }, + { + "batch_loss": 1.8886240720748901, + "innovation_rms": 0.0016107338134362537, + "instruction_rms": 0.0016107337980926072, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334159567952156, + "max_absolute_post_projection_soma_slope": 1.0751787016261005e-08, + "max_absolute_pre_projection_soma_slope": 0.029334159567952156, + "max_positive_post_projection_soma_slope": 6.686572628922249e-09, + "min_post_projection_soma_slope": -1.0751787016261005e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8520656481923658e-08, + "pre_projection_traffic_rms_ratio": 0.004809591744199306 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.0876970291137695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.77730560302734, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10646075755357742, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6778262853622437, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21344465017318726, + "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": 0.5014038681983948, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11773309111595154, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7753864526748657, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10795164853334427, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10659022629261017, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6677860021591187, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2133585512638092, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1754039525985718, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019254906738898677, + "step": 82, + "teaching_rms": 0.0016107338134362537, + "traffic_rms": 0.019188998669563456 + }, + { + "batch_loss": 1.8732566833496094, + "innovation_rms": 0.0016014751710566912, + "instruction_rms": 0.0016014751725999263, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 6.976392796786968e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 6.976392796786968e-09, + "min_post_projection_soma_slope": -3.4843741225643043e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8391913762405706e-08, + "pre_projection_traffic_rms_ratio": 0.004713027279221265 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.103570938110352, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.15056610107422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1104368194937706, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6822429299354553, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22309610247612, + "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": 0.4936787486076355, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10782255232334137, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7752361297607422, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09862232953310013, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11056627333164215, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6786320805549622, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2230580747127533, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1763252019882202, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01926794569521696, + "step": 83, + "teaching_rms": 0.0016014751710566912, + "traffic_rms": 0.019203573002761008 + }, + { + "batch_loss": 1.8909169435501099, + "innovation_rms": 0.0015873003586116495, + "instruction_rms": 0.0015873003523835833, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 5.600536479022367e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 5.600536479022367e-09, + "min_post_projection_soma_slope": -3.4429197270924305e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8584768030217698e-08, + "pre_projection_traffic_rms_ratio": 0.004743129622412476 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.121748924255371, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.77092742919922, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08980316668748856, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6931744813919067, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14687736332416534, + "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": 0.49355176091194153, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0709204152226448, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7763727903366089, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1183663159608841, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08993260562419891, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6895853281021118, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1468019187450409, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1775377988815308, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019221280161215333, + "step": 84, + "teaching_rms": 0.0015873003586116495, + "traffic_rms": 0.019159473262767866 + }, + { + "batch_loss": 2.1137661933898926, + "innovation_rms": 0.0017414411432165535, + "instruction_rms": 0.0017414411432165535, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 1.40839810924831e-08, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 1.40839810924831e-08, + "min_post_projection_soma_slope": -9.0585237089158e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8671423658138732e-08, + "pre_projection_traffic_rms_ratio": 0.004605561378135786 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.142175197601318, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.14471435546875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07210607826709747, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.70713210105896, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19493180513381958, + "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": 0.4925718307495117, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07053199410438538, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.778085708618164, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10636095702648163, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11438792943954468, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.683107316493988, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19489651918411255, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1786421537399292, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019214077247858907, + "step": 85, + "teaching_rms": 0.0017414411432165535, + "traffic_rms": 0.01913930090636705 + }, + { + "batch_loss": 1.9396263360977173, + "innovation_rms": 0.0016036496426800456, + "instruction_rms": 0.0016036496426800456, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 6.500827876010362e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.792497065864154e-09, + "min_post_projection_soma_slope": -6.500827876010362e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8466557826929388e-08, + "pre_projection_traffic_rms_ratio": 0.004864645293526896 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.155729293823242, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.1197509765625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06162011995911598, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7220757603645325, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22500255703926086, + "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": 0.49442803859710693, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10401079058647156, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7784082889556885, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0926239937543869, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1038842424750328, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6605984568595886, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22508864104747772, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1792141199111938, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019295666395816823, + "step": 86, + "teaching_rms": 0.0016036496426800456, + "traffic_rms": 0.019232498756409054 + }, + { + "batch_loss": 1.8598785400390625, + "innovation_rms": 0.0016422436187510888, + "instruction_rms": 0.001642243627780637, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 9.951126145324451e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 8.312371235774663e-09, + "min_post_projection_soma_slope": -9.951126145324451e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8516107235093703e-08, + "pre_projection_traffic_rms_ratio": 0.004848735706901542 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.17455530166626, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.33753967285156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06308972090482712, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7389966249465942, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25711360573768616, + "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": 0.49534496665000916, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09535928815603256, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7739145755767822, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1224101334810257, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10552506148815155, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6470754146575928, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25715163350105286, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1796103715896606, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01920972590608001, + "step": 87, + "teaching_rms": 0.0016422436187510888, + "traffic_rms": 0.019139287302285604 + }, + { + "batch_loss": 1.9113476276397705, + "innovation_rms": 0.0014768233109348529, + "instruction_rms": 0.0014768233142818381, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 7.767850362938589e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.938457115992833e-09, + "min_post_projection_soma_slope": -7.767850362938589e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8479015177410354e-08, + "pre_projection_traffic_rms_ratio": 0.004347425702641059 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.187727451324463, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.54716491699219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06294352561235428, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7560596466064453, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32302623987197876, + "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": 0.49683213233947754, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09916726499795914, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.772418737411499, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10003061592578888, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11234866082668304, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6371924877166748, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32306426763534546, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1801202297210693, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01922171540134791, + "step": 88, + "teaching_rms": 0.0014768233109348529, + "traffic_rms": 0.0191641365388655 + }, + { + "batch_loss": 1.930641531944275, + "innovation_rms": 0.0014253316493997633, + "instruction_rms": 0.001425331652867662, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 2.952078137496983e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 2.952078137496983e-09, + "min_post_projection_soma_slope": -1.9320089972296728e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8663600380662446e-08, + "pre_projection_traffic_rms_ratio": 0.004587292971429906 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.2070465087890625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.84176635742188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05999215692281723, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7715378403663635, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2548442780971527, + "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": 0.4969886839389801, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10552532225847244, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7704620361328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08640116453170776, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11479468643665314, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6307229995727539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.254882276058197, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1806325912475586, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01920652758702903, + "step": 89, + "teaching_rms": 0.0014253316493997633, + "traffic_rms": 0.019152751308177646 + }, + { + "batch_loss": 1.886573076248169, + "innovation_rms": 0.0014747439800332461, + "instruction_rms": 0.0014747439833849504, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 1.0694657603949054e-08, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 1.0694657603949054e-08, + "min_post_projection_soma_slope": -6.130003171733733e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.838301958571513e-08, + "pre_projection_traffic_rms_ratio": 0.004796929761414933 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.226022720336914, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.8359375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05840062350034714, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7843760251998901, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22116829454898834, + "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": 0.4983760714530945, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0973447933793068, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.764394760131836, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08529168367385864, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10046832263469696, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6297003626823425, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2210538685321808, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1810945272445679, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019274227054845557, + "step": 90, + "teaching_rms": 0.0014747439800332461, + "traffic_rms": 0.019212601687986903 + }, + { + "batch_loss": 1.9399633407592773, + "innovation_rms": 0.0014155499685506354, + "instruction_rms": 0.001415549977280292, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.573808969927541e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.573808969927541e-09, + "min_post_projection_soma_slope": -3.375075108280612e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8509321746342363e-08, + "pre_projection_traffic_rms_ratio": 0.004768779023095412 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.24439811706543, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.30764770507812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04787992686033249, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7926287651062012, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22346919775009155, + "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": 0.4951139986515045, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07821579277515411, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7565429210662842, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08879014104604721, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10530619323253632, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6389745473861694, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2233547866344452, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1812626123428345, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019286971913963454, + "step": 91, + "teaching_rms": 0.0014155499685506354, + "traffic_rms": 0.019233221007282038 + }, + { + "batch_loss": 1.9246881008148193, + "innovation_rms": 0.00158003181060419, + "instruction_rms": 0.0015800318137325485, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 2.5625830346598377e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 2.4192869929606786e-09, + "min_post_projection_soma_slope": -2.5625830346598377e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.863222910504983e-08, + "pre_projection_traffic_rms_ratio": 0.005061644609165231 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.262993335723877, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.74143981933594, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04572674259543419, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7995656132698059, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42183536291122437, + "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": 0.4983731508255005, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09993943572044373, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7549258470535278, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07742145657539368, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07294246554374695, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6424025893211365, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.42179733514785767, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1814264059066772, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019266754542965316, + "step": 92, + "teaching_rms": 0.00158003181060419, + "traffic_rms": 0.019202340788075953 + }, + { + "batch_loss": 2.132589817047119, + "innovation_rms": 0.0016194896299611989, + "instruction_rms": 0.0016194896299611989, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 1.2497703316682873e-08, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 1.2497703316682873e-08, + "min_post_projection_soma_slope": -3.6703322603415245e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.854559802473976e-08, + "pre_projection_traffic_rms_ratio": 0.004687793297680208 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.2779130935668945, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.74720001220703, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04972648620605469, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8008801341056824, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4277839958667755, + "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": 0.5002182126045227, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12999661266803741, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.754652500152588, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07001884281635284, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04971405491232872, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6324452757835388, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.4277459979057312, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1811015605926514, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01924422061387785, + "step": 93, + "teaching_rms": 0.0016194896299611989, + "traffic_rms": 0.01918184169837474 + }, + { + "batch_loss": 1.8856747150421143, + "innovation_rms": 0.0014950255664086607, + "instruction_rms": 0.0014950255697148956, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.540099046186242e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 5.540099046186242e-09, + "min_post_projection_soma_slope": -2.5411162063448955e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8497376314822603e-08, + "pre_projection_traffic_rms_ratio": 0.004771441357680828 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.293961048126221, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.8565444946289, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0404844731092453, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8013449907302856, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3623983860015869, + "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": 0.49752455949783325, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12058661133050919, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.755583643913269, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07731199264526367, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07585226744413376, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6140018105506897, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.3623603880405426, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1805531978607178, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01921507238502464, + "step": 94, + "teaching_rms": 0.0014950255664086607, + "traffic_rms": 0.0191559059191593 + }, + { + "batch_loss": 1.9451792240142822, + "innovation_rms": 0.0016598829079024803, + "instruction_rms": 0.0016598828974799553, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.0403658196908054e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 4.0403658196908054e-09, + "min_post_projection_soma_slope": -3.6489562482699966e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.838190660396902e-08, + "pre_projection_traffic_rms_ratio": 0.004896147222318753 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.308791160583496, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.83032989501953, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.042360734194517136, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8013303279876709, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23105065524578094, + "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": 0.4915270209312439, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11324521154165268, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7540208101272583, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0689050555229187, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06928904354572296, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.5981788635253906, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23115533590316772, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1802436113357544, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01927058642083657, + "step": 95, + "teaching_rms": 0.0016598829079024803, + "traffic_rms": 0.01920038137780454 + }, + { + "batch_loss": 2.1411492824554443, + "innovation_rms": 0.0017368724425636368, + "instruction_rms": 0.0017368724454095027, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 7.689953562817209e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 6.612089098467777e-09, + "min_post_projection_soma_slope": -7.689953562817209e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8677412638657414e-08, + "pre_projection_traffic_rms_ratio": 0.004896482802938464 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.322996139526367, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.38351440429688, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.061034977436065674, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8012177348136902, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1631021499633789, + "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": 0.4805263578891754, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1702573299407959, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.753466248512268, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10413114726543427, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07464124262332916, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6042633652687073, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16316939890384674, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1800036430358887, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01923239435639311, + "step": 96, + "teaching_rms": 0.0017368724425636368, + "traffic_rms": 0.019159511712165067 + }, + { + "batch_loss": 1.882181167602539, + "innovation_rms": 0.0017314098701028218, + "instruction_rms": 0.0017314098701028218, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 1.0583497633831485e-08, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 5.488355547811352e-09, + "min_post_projection_soma_slope": -1.0583497633831485e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8435684474514427e-08, + "pre_projection_traffic_rms_ratio": 0.004909877965060126 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.333898067474365, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.46809387207031, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06031368300318718, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8003883957862854, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19996196031570435, + "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": 0.481301486492157, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15137214958667755, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7544121742248535, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09410949051380157, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06976568698883057, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6183092594146729, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2000233232975006, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.179762363433838, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01928927205642363, + "step": 97, + "teaching_rms": 0.0017314098701028218, + "traffic_rms": 0.019211789462465205 + }, + { + "batch_loss": 1.930335521697998, + "innovation_rms": 0.0015622203930843372, + "instruction_rms": 0.0015622203915023243, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.4168587959015895e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.4168587959015895e-09, + "min_post_projection_soma_slope": -3.3258389375845354e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.855012987241388e-08, + "pre_projection_traffic_rms_ratio": 0.004768433474867547 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.346580982208252, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.8772964477539, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.057516422122716904, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8012938499450684, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24377571046352386, + "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": 0.4816180169582367, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10961709171533585, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7537708282470703, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07959406822919846, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07266667485237122, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6318917870521545, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24383705854415894, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1791867017745972, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019246403362694613, + "step": 98, + "teaching_rms": 0.0015622203930843372, + "traffic_rms": 0.019182340620802315 + }, + { + "batch_loss": 2.0448975563049316, + "innovation_rms": 0.0015757203950981509, + "instruction_rms": 0.0015757203950981509, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 1.5233538874781516e-08, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 1.5233538874781516e-08, + "min_post_projection_soma_slope": -4.689094001975036e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8485587673200073e-08, + "pre_projection_traffic_rms_ratio": 0.004669580367158135 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.353863716125488, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.64274597167969, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04942707344889641, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8020290732383728, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18733251094818115, + "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": 0.4825192987918854, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08007992804050446, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7509788274765015, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07446011900901794, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06360308080911636, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6412390470504761, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18739384412765503, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1785252094268799, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019255306919275992, + "step": 99, + "teaching_rms": 0.0015757203950981509, + "traffic_rms": 0.0191908113882066 + }, + { + "batch_loss": 1.869093418121338, + "innovation_rms": 0.001376484929645537, + "instruction_rms": 0.001376484929645537, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 8.719117872146853e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.002915027749964e-09, + "min_post_projection_soma_slope": -8.719117872146853e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.845450615403015e-08, + "pre_projection_traffic_rms_ratio": 0.004662200782210509 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.363497257232666, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.4009780883789, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047922153025865555, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8034887909889221, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1347682774066925, + "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": 0.4822424650192261, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09798623621463776, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7493623495101929, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09987150132656097, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.055153194814920425, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6507457494735718, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13463592529296875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1777878999710083, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019246139306445368, + "step": 100, + "teaching_rms": 0.001376484929645537, + "traffic_rms": 0.019195925913329918 + }, + { + "batch_loss": 1.7942994832992554, + "innovation_rms": 0.0014415796118663388, + "instruction_rms": 0.0014415796101519326, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 1.0786592952172214e-08, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 6.21889650886942e-09, + "min_post_projection_soma_slope": -1.0786592952172214e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8549853761832372e-08, + "pre_projection_traffic_rms_ratio": 0.0051014685139739845 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.384062767028809, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.9776611328125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05227866768836975, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8030678629875183, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14726021885871887, + "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": 0.4799225628376007, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08661314100027084, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7457259893417358, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11277121305465698, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07076027244329453, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6568228602409363, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1471278816461563, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1769307851791382, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01931133689905045, + "step": 101, + "teaching_rms": 0.0014415796118663388, + "traffic_rms": 0.019256302589171947 + }, + { + "batch_loss": 1.8939592838287354, + "innovation_rms": 0.0014039936629867843, + "instruction_rms": 0.0014039936489043674, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 4.528737385101067e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.528737385101067e-09, + "min_post_projection_soma_slope": -4.397498365449337e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8550169992726735e-08, + "pre_projection_traffic_rms_ratio": 0.005025922822249097 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.392218589782715, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.8664779663086, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.049075350165367126, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8032291531562805, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19983047246932983, + "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": 0.4771595895290375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0703049972653389, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.744888424873352, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09883177280426025, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04937915503978729, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6539771556854248, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1996685415506363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1761653423309326, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01932193583843706, + "step": 102, + "teaching_rms": 0.0014039936629867843, + "traffic_rms": 0.01926895449752263 + }, + { + "batch_loss": 1.946509599685669, + "innovation_rms": 0.0015593380575013704, + "instruction_rms": 0.0015593380479917466, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 7.913886435062523e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 5.9733009649676205e-09, + "min_post_projection_soma_slope": -7.913886435062523e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8601538774236218e-08, + "pre_projection_traffic_rms_ratio": 0.005039715748921872 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.398850440979004, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.95849609375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.037292104214429855, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.805267870426178, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1531570702791214, + "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": 0.4772689938545227, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09399692714214325, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7423454523086548, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09955734759569168, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.083330437541008, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6534499526023865, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15337052941322327, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.175371766090393, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01928768776223919, + "step": 103, + "teaching_rms": 0.0015593380575013704, + "traffic_rms": 0.019221341150767415 + }, + { + "batch_loss": 1.9186862707138062, + "innovation_rms": 0.0014212519851350403, + "instruction_rms": 0.001421251992090747, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 3.5875136195073765e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 2.890685246725866e-09, + "min_post_projection_soma_slope": -3.5875136195073765e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8415417043275308e-08, + "pre_projection_traffic_rms_ratio": 0.005104252966917264 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.407721996307373, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.72028350830078, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03626447170972824, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8092426657676697, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2654809355735779, + "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": 0.47888195514678955, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11798569560050964, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7411482334136963, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08997323364019394, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07013686001300812, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6465940475463867, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2658001780509949, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1744885444641113, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019362630166039292, + "step": 104, + "teaching_rms": 0.0014212519851350403, + "traffic_rms": 0.019307965425060795 + }, + { + "batch_loss": 1.8309667110443115, + "innovation_rms": 0.001389053277257496, + "instruction_rms": 0.0013890532665820825, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 6.598067425755971e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 6.598067425755971e-09, + "min_post_projection_soma_slope": -4.190777946888602e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8468959693370178e-08, + "pre_projection_traffic_rms_ratio": 0.004949367221790108 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.417497634887695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.1572265625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03648222237825394, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8150235414505005, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32705020904541016, + "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": 0.4811766445636749, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13212260603904724, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7439897060394287, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08955849707126617, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.061419010162353516, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6392917037010193, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.32736945152282715, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1734589338302612, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019357995126799407, + "step": 105, + "teaching_rms": 0.001389053277257496, + "traffic_rms": 0.019306197190647474 + }, + { + "batch_loss": 1.8591158390045166, + "innovation_rms": 0.0014876108538248443, + "instruction_rms": 0.00148761085050213, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.377901235448235e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.377901235448235e-09, + "min_post_projection_soma_slope": -4.0932501832458e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.841373972408984e-08, + "pre_projection_traffic_rms_ratio": 0.00497988298510509 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.426308631896973, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.66656494140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05331892520189285, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8180934190750122, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35450220108032227, + "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": 0.4827594757080078, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10780730843544006, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7444132566452026, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08051519095897675, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05347052589058876, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.634113073348999, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35482144355773926, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1724687814712524, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01942241943859898, + "step": 106, + "teaching_rms": 0.0014876108538248443, + "traffic_rms": 0.019365556673896382 + }, + { + "batch_loss": 1.8503642082214355, + "innovation_rms": 0.0015585951979719078, + "instruction_rms": 0.0015585951979719078, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675612226128578, + "max_absolute_post_projection_soma_slope": 1.0012998430397602e-08, + "max_absolute_pre_projection_soma_slope": 0.029675612226128578, + "max_positive_post_projection_soma_slope": 1.0012998430397602e-08, + "min_post_projection_soma_slope": -4.2724250803871655e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8516110912669223e-08, + "pre_projection_traffic_rms_ratio": 0.005324573220898355 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.443514823913574, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.02980041503906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.035071901977062225, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8219265937805176, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2652084231376648, + "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": 0.4849427044391632, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07903508841991425, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7450112104415894, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08081842958927155, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05583903193473816, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.637017011642456, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.265527606010437, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1714611053466797, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01945726661831598, + "step": 107, + "teaching_rms": 0.0015585951979719078, + "traffic_rms": 0.019393078270516947 + }, + { + "batch_loss": 1.8499635457992554, + "innovation_rms": 0.0015508977907895479, + "instruction_rms": 0.0015508977939766733, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 7.0946888364176175e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.834886630471601e-09, + "min_post_projection_soma_slope": -7.0946888364176175e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.866491115334533e-08, + "pre_projection_traffic_rms_ratio": 0.005085634071006991 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.461606979370117, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.40342712402344, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04228513687849045, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8231053352355957, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1530730277299881, + "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": 0.48815402388572693, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08292239904403687, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7428752183914185, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09037080407142639, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04508672654628754, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6414276957511902, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1531343162059784, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1703088283538818, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019429340444394937, + "step": 108, + "teaching_rms": 0.0015508977907895479, + "traffic_rms": 0.019367697534737875 + }, + { + "batch_loss": 1.7900118827819824, + "innovation_rms": 0.0015587713901834565, + "instruction_rms": 0.0015587713933544833, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 4.911343776115018e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 4.911343776115018e-09, + "min_post_projection_soma_slope": -4.127951314103484e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8520841488856443e-08, + "pre_projection_traffic_rms_ratio": 0.0052074035702345 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.473299980163574, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.78761291503906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04639538377523422, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.820736825466156, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14882168173789978, + "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": 0.492798388004303, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07646603882312775, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7429007291793823, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09852932393550873, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.046430181711912155, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6563012599945068, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14873579144477844, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.169191837310791, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01949482163044397, + "step": 109, + "teaching_rms": 0.0015587713901834565, + "traffic_rms": 0.019435041235814986 + }, + { + "batch_loss": 1.9379318952560425, + "innovation_rms": 0.001721009597254553, + "instruction_rms": 0.001721009597254553, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 3.958778638235572e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.958778638235572e-09, + "min_post_projection_soma_slope": -3.5245004692541215e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8514205756653158e-08, + "pre_projection_traffic_rms_ratio": 0.0051689931957223725 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.486965656280518, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.96776580810547, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0455295667052269, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8173328042030334, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.372723251581192, + "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": 0.49446022510528564, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10770449042320251, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.741453766822815, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13173405826091766, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07750590145587921, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6729122996330261, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.37240415811538696, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1680876016616821, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019491242724928918, + "step": 110, + "teaching_rms": 0.001721009597254553, + "traffic_rms": 0.019418838085734556 + }, + { + "batch_loss": 1.8632315397262573, + "innovation_rms": 0.0017436604674089876, + "instruction_rms": 0.0017436604702437744, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.9060586704285924e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.523000779992458e-09, + "min_post_projection_soma_slope": -4.9060586704285924e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8668051811744326e-08, + "pre_projection_traffic_rms_ratio": 0.005064920036717966 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.502723693847656, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.20765686035156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06389039754867554, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8116081953048706, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35764628648757935, + "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": 0.49518883228302, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10617377609014511, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7331268787384033, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1748771071434021, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09634831547737122, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6917902827262878, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.35732725262641907, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.167110800743103, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019472899487619044, + "step": 111, + "teaching_rms": 0.0017436604674089876, + "traffic_rms": 0.019394648518994865 + }, + { + "batch_loss": 1.867575764656067, + "innovation_rms": 0.0014523427436591773, + "instruction_rms": 0.0014523427283438689, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.090628807380426e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.7033358601945565e-09, + "min_post_projection_soma_slope": -5.090628807380426e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.835792721632719e-08, + "pre_projection_traffic_rms_ratio": 0.005099271924818519 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.504007816314697, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.36076354980469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06944125145673752, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8068146109580994, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21273234486579895, + "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": 0.4984797239303589, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06850261241197586, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7258222103118896, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1583445519208908, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12273694574832916, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7021898031234741, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21241332590579987, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1662137508392334, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019531710895981023, + "step": 112, + "teaching_rms": 0.0014523427436591773, + "traffic_rms": 0.019477533064204713 + }, + { + "batch_loss": 1.7624340057373047, + "innovation_rms": 0.0016004908430924394, + "instruction_rms": 0.0016004908322831486, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.217109493888984e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 5.59398172228498e-09, + "min_post_projection_soma_slope": -6.217109493888984e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8676617272841002e-08, + "pre_projection_traffic_rms_ratio": 0.004914593637810303 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.511519432067871, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.20809173583984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07648694515228271, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8007266521453857, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16252440214157104, + "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": 0.49964961409568787, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08005227893590927, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7229598760604858, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12999992072582245, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0763479471206665, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7034389972686768, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16220542788505554, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165490984916687, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01949440681873736, + "step": 113, + "teaching_rms": 0.0016004908430924394, + "traffic_rms": 0.019428402118974748 + }, + { + "batch_loss": 2.0031328201293945, + "innovation_rms": 0.0017871570681117714, + "instruction_rms": 0.0017871570750262532, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.961239584768464e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 6.961239584768464e-09, + "min_post_projection_soma_slope": -3.6982652495964885e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.848344959490409e-08, + "pre_projection_traffic_rms_ratio": 0.0053693384256764485 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.518121242523193, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.0888671875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0741046667098999, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7957408428192139, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14518073201179504, + "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": 0.4957079291343689, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09199807047843933, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7228676080703735, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1226322278380394, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07415688037872314, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7024210691452026, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14522194862365723, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1648629903793335, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019592492571525243, + "step": 114, + "teaching_rms": 0.0017871570681117714, + "traffic_rms": 0.019513311865399948 + }, + { + "batch_loss": 1.7503184080123901, + "innovation_rms": 0.0014729531864622844, + "instruction_rms": 0.0014729531864622844, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 3.990371588713515e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 2.4110391461107383e-09, + "min_post_projection_soma_slope": -3.990371588713515e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.863279757974016e-08, + "pre_projection_traffic_rms_ratio": 0.0049676340999993485 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.524833679199219, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.08949279785156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06698695570230484, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7909932136535645, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15575607120990753, + "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": 0.4917697012424469, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09177317470312119, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7220509052276611, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09290692210197449, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06684798747301102, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7007650136947632, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15579728782176971, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1643970012664795, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019499635336238923, + "step": 115, + "teaching_rms": 0.0014729531864622844, + "traffic_rms": 0.01944644385351386 + }, + { + "batch_loss": 1.7545950412750244, + "innovation_rms": 0.001443813975136686, + "instruction_rms": 0.0014438139802719452, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 5.210733622362795e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 5.210733622362795e-09, + "min_post_projection_soma_slope": -4.522801688722211e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8492184090991125e-08, + "pre_projection_traffic_rms_ratio": 0.005290508345318563 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.52907657623291, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.43109130859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0625663697719574, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7886077761650085, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14039361476898193, + "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": 0.49076083302497864, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06386157870292664, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7192384004592896, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08619199693202972, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06874595582485199, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7013483047485352, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1406078040599823, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1642037630081177, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019601964870759245, + "step": 116, + "teaching_rms": 0.001443813975136686, + "traffic_rms": 0.0195470996769465 + }, + { + "batch_loss": 1.7775027751922607, + "innovation_rms": 0.0017767354923862607, + "instruction_rms": 0.0017767354854312215, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 7.433675897061676e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 7.433675897061676e-09, + "min_post_projection_soma_slope": -3.278883164981039e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8561568081977774e-08, + "pre_projection_traffic_rms_ratio": 0.005361823125527708 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.537374019622803, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.37590026855469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.055643096566200256, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7882022857666016, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1625848412513733, + "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": 0.4891521632671356, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08287444710731506, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7155311107635498, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08065475523471832, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0809759795665741, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7049232125282288, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16271758079528809, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1639080047607422, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019607763098800617, + "step": 117, + "teaching_rms": 0.0017767354923862607, + "traffic_rms": 0.019526890516872942 + }, + { + "batch_loss": 1.844988465309143, + "innovation_rms": 0.0017191281079899816, + "instruction_rms": 0.0017191280993642615, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.819093568642074e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 3.7444700673461284e-09, + "min_post_projection_soma_slope": -4.819093568642074e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8531521197934715e-08, + "pre_projection_traffic_rms_ratio": 0.005224212046405954 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.546329975128174, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.54061126708984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04977036267518997, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7894780039787292, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16296041011810303, + "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": 0.48849043250083923, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07028172165155411, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7122564315795898, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1192612498998642, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08818746358156204, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7110928893089294, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16309258341789246, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1636443138122559, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019576234092535807, + "step": 118, + "teaching_rms": 0.0017191281079899816, + "traffic_rms": 0.019501450313696322 + }, + { + "batch_loss": 1.8747748136520386, + "innovation_rms": 0.0016949320394134665, + "instruction_rms": 0.001694932045246038, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.2493555341138745e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.2493555341138745e-09, + "min_post_projection_soma_slope": -4.218314586523775e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.854096785858787e-08, + "pre_projection_traffic_rms_ratio": 0.005387172165175788 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.554376125335693, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.81082153320312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.053735461086034775, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.790067195892334, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18867270648479462, + "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": 0.49520936608314514, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10097330808639526, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7105382680892944, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11128751933574677, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12056618183851242, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7155900001525879, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18880486488342285, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1634681224822998, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019626703073559168, + "step": 119, + "teaching_rms": 0.0016949320394134665, + "traffic_rms": 0.019555478187081348 + }, + { + "batch_loss": 1.8754780292510986, + "innovation_rms": 0.0018792990164005212, + "instruction_rms": 0.0018792990164005212, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 1.1287495382816815e-08, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 1.1287495382816815e-08, + "min_post_projection_soma_slope": -8.43938074979178e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8625812051116434e-08, + "pre_projection_traffic_rms_ratio": 0.0053252078613672356 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.5603766441345215, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.12689971923828, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0522637814283371, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7903639078140259, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15081468224525452, + "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": 0.4968405067920685, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08866307884454727, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7082282304763794, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12906664609909058, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10178376734256744, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7202333211898804, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15094681084156036, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1633766889572144, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019645950163503078, + "step": 120, + "teaching_rms": 0.0018792990164005212, + "traffic_rms": 0.019556124587036712 + }, + { + "batch_loss": 1.8114283084869385, + "innovation_rms": 0.0018402841638182838, + "instruction_rms": 0.0018402841745620722, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 8.36154612215978e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 8.36154612215978e-09, + "min_post_projection_soma_slope": -5.515883749751538e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.870858730121436e-08, + "pre_projection_traffic_rms_ratio": 0.005181257394867806 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.562533855438232, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.43010711669922, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04445655643939972, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7894110083580017, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18441948294639587, + "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": 0.49533456563949585, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07890037447214127, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.707529067993164, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12850694358348846, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09096566587686539, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7184786200523376, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18464702367782593, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1630859375, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019588795737510736, + "step": 121, + "teaching_rms": 0.0018402841638182838, + "traffic_rms": 0.019502836392485002 + }, + { + "batch_loss": 1.8606983423233032, + "innovation_rms": 0.002031602075977839, + "instruction_rms": 0.0020316020650292988, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.095300543127678e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.066914804923272e-09, + "min_post_projection_soma_slope": -4.095300543127678e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.878674676160739e-08, + "pre_projection_traffic_rms_ratio": 0.005154718180029104 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.569750785827637, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.52783966064453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.048753321170806885, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7850175499916077, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20610225200653076, + "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": 0.4998314678668976, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12166231870651245, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7051399946212769, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11025087535381317, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1557663232088089, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7184886336326599, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20578356087207794, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1631484031677246, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019604586283157283, + "step": 122, + "teaching_rms": 0.002031602075977839, + "traffic_rms": 0.019500313460231813 + }, + { + "batch_loss": 1.8526735305786133, + "innovation_rms": 0.0018589440089241086, + "instruction_rms": 0.0018589440115830945, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.7629842292451485e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.464125069603142e-09, + "min_post_projection_soma_slope": -4.7629842292451485e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8478198664920535e-08, + "pre_projection_traffic_rms_ratio": 0.005075070885688909 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.571298599243164, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.75340270996094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05282676964998245, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7807754874229431, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.28702542185783386, + "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": 0.509884238243103, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1857490837574005, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7034069299697876, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16823118925094604, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17178528010845184, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7201335430145264, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2867067754268646, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1635596752166748, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019609265554584164, + "step": 123, + "teaching_rms": 0.0018589440089241086, + "traffic_rms": 0.01952043805736142 + }, + { + "batch_loss": 1.8824667930603027, + "innovation_rms": 0.0020839687398424077, + "instruction_rms": 0.0020839687208674355, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 7.364047149849284e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 7.364047149849284e-09, + "min_post_projection_soma_slope": -4.788086371831923e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.86881696399464e-08, + "pre_projection_traffic_rms_ratio": 0.004837260051050626 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.583384037017822, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.0363998413086, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0740249902009964, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7765114903450012, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21291396021842957, + "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": 0.5219914317131042, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22999566793441772, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7036720514297485, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1863296777009964, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1559228003025055, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7193971872329712, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21259534358978271, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1639902591705322, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019647611272180152, + "step": 124, + "teaching_rms": 0.0020839687398424077, + "traffic_rms": 0.019535366374405578 + }, + { + "batch_loss": 1.9241138696670532, + "innovation_rms": 0.002112555425889737, + "instruction_rms": 0.002112555425889737, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 4.562991318124432e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.50500436954826e-09, + "min_post_projection_soma_slope": -4.562991318124432e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8485935564142598e-08, + "pre_projection_traffic_rms_ratio": 0.005075986027332739 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.599587440490723, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.95916748046875, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07438965141773224, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7731072306632996, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2564009130001068, + "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": 0.5325573682785034, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22361557185649872, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7027804851531982, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17983880639076233, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15754938125610352, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7215815782546997, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2562965452671051, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1644810438156128, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01969482887770723, + "step": 125, + "teaching_rms": 0.002112555425889737, + "traffic_rms": 0.01958190119962354 + }, + { + "batch_loss": 1.895248293876648, + "innovation_rms": 0.0018029606889212488, + "instruction_rms": 0.0018029606916627982, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 1.1968263713413307e-08, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 1.1968263713413307e-08, + "min_post_projection_soma_slope": -8.341989321536403e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8508586524095245e-08, + "pre_projection_traffic_rms_ratio": 0.004975505358852176 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.620161533355713, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.99722290039062, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0678328201174736, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7721885442733765, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19569115340709686, + "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": 0.5362944602966309, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13706737756729126, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.700932264328003, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14962449669837952, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13843370974063873, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.721280038356781, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19558680057525635, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.164929747581482, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019703396498952197, + "step": 126, + "teaching_rms": 0.0018029606889212488, + "traffic_rms": 0.019618174451027696 + }, + { + "batch_loss": 1.8663088083267212, + "innovation_rms": 0.0018078189311907667, + "instruction_rms": 0.0018078189257224027, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 3.728141795278361e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.728141795278361e-09, + "min_post_projection_soma_slope": -2.7079791742323778e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8481774363740472e-08, + "pre_projection_traffic_rms_ratio": 0.005080586543424925 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.644413471221924, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.06401824951172, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.057931654155254364, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.773159384727478, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10903146862983704, + "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": 0.5379701852798462, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13424739241600037, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.701801061630249, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10292837023735046, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12238013744354248, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7266073822975159, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10916349291801453, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1654914617538452, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01972755893995442, + "step": 127, + "teaching_rms": 0.0018078189311907667, + "traffic_rms": 0.01963971522647071 + }, + { + "batch_loss": 1.6424249410629272, + "innovation_rms": 0.001589790503826419, + "instruction_rms": 0.0015897905131538855, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.560481325910359e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.560481325910359e-09, + "min_post_projection_soma_slope": -4.4356247563825946e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8637981667950305e-08, + "pre_projection_traffic_rms_ratio": 0.004785089186524601 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.678128242492676, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.14265441894531, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0504281222820282, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7744125723838806, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11311113834381104, + "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": 0.5421188473701477, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15392716228961945, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7139333486557007, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12132246047258377, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12186244875192642, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7365622520446777, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11324404180049896, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1661192178726196, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01967163466932308, + "step": 128, + "teaching_rms": 0.001589790503826419, + "traffic_rms": 0.019601470430818363 + }, + { + "batch_loss": 1.9094597101211548, + "innovation_rms": 0.0020375568206674024, + "instruction_rms": 0.0020375568036861127, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.911186651275102e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.3803187743662875e-09, + "min_post_projection_soma_slope": -5.911186651275102e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8679341339057797e-08, + "pre_projection_traffic_rms_ratio": 0.004930708055732851 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.714150905609131, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.92771911621094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05808223783969879, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.776419997215271, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17869603633880615, + "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": 0.5414724946022034, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20805387198925018, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.727658748626709, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1372535228729248, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09267211705446243, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7518446445465088, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17892339825630188, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1665911674499512, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019750521415273666, + "step": 129, + "teaching_rms": 0.0020375568206674024, + "traffic_rms": 0.019644458717214545 + }, + { + "batch_loss": 1.7091357707977295, + "innovation_rms": 0.0019049628655208664, + "instruction_rms": 0.0019049628590339866, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.6034958895168074e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.832281241817782e-09, + "min_post_projection_soma_slope": -5.6034958895168074e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8794337414659458e-08, + "pre_projection_traffic_rms_ratio": 0.004913345861341818 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.7539963722229, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.740478515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07691391557455063, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.777292013168335, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15734921395778656, + "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": 0.5587054491043091, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19910494983196259, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.744649052619934, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1699027270078659, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11674922704696655, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7650055289268494, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1575765460729599, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1669591665267944, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019743941523584854, + "step": 130, + "teaching_rms": 0.0019049628655208664, + "traffic_rms": 0.019649699462571408 + }, + { + "batch_loss": 1.8225204944610596, + "innovation_rms": 0.001887132666448856, + "instruction_rms": 0.0018871326716873918, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 6.781479378048516e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 6.781479378048516e-09, + "min_post_projection_soma_slope": -2.270972521145609e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8621118849701134e-08, + "pre_projection_traffic_rms_ratio": 0.005018641829620039 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.794708728790283, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.37086486816406, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0835895836353302, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7794424295425415, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24588501453399658, + "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": 0.5760838389396667, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17378368973731995, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7606523036956787, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16003277897834778, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0995040312409401, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7688807845115662, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24620340764522552, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1675152778625488, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01976848835715474, + "step": 131, + "teaching_rms": 0.001887132666448856, + "traffic_rms": 0.01967737178275219 + }, + { + "batch_loss": 1.9757412672042847, + "innovation_rms": 0.0018769363369450626, + "instruction_rms": 0.0018769363343115657, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.914151225359319e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.914151225359319e-09, + "min_post_projection_soma_slope": -2.911417551487716e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8595499537827665e-08, + "pre_projection_traffic_rms_ratio": 0.00470029650663245 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.82988166809082, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.84488677978516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09210305660963058, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.780180811882019, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2523738443851471, + "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": 0.5863209962844849, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13201287388801575, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7756685018539429, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15016154944896698, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10206900537014008, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7719417214393616, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25269222259521484, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1676054000854492, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01978584785438417, + "step": 132, + "teaching_rms": 0.0018769363369450626, + "traffic_rms": 0.01969755171601145 + }, + { + "batch_loss": 1.8441319465637207, + "innovation_rms": 0.0018599846282276603, + "instruction_rms": 0.0018599846229126638, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 6.820312759003855e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.131343267488319e-09, + "min_post_projection_soma_slope": -6.820312759003855e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8778273490340378e-08, + "pre_projection_traffic_rms_ratio": 0.005079103141561191 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.874258518218994, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.25800323486328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08786925673484802, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7799439430236816, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18233808875083923, + "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": 0.5897104740142822, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09823904186487198, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7814717292785645, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0863656997680664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14022332429885864, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7615007162094116, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1826564371585846, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1675258874893188, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019781657528904862, + "step": 133, + "teaching_rms": 0.0018599846282276603, + "traffic_rms": 0.019699633757529016 + }, + { + "batch_loss": 1.828893780708313, + "innovation_rms": 0.0017453442209489564, + "instruction_rms": 0.0017453442181169041, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 6.397819163339591e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 6.397819163339591e-09, + "min_post_projection_soma_slope": -4.571373501960352e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8638474928721487e-08, + "pre_projection_traffic_rms_ratio": 0.005217163438794801 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.915832996368408, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.50418853759766, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08526696264743805, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7812525033950806, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17863653600215912, + "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": 0.5969672203063965, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13687998056411743, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7827372550964355, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10648586601018906, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1467103213071823, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7460166811943054, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17883475124835968, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1671457290649414, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019717330523161616, + "step": 134, + "teaching_rms": 0.0017453442209489564, + "traffic_rms": 0.019642611415618835 + }, + { + "batch_loss": 1.9314861297607422, + "innovation_rms": 0.0017763656391565899, + "instruction_rms": 0.0017763656391565899, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 7.829993542429747e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 7.829993542429747e-09, + "min_post_projection_soma_slope": -5.225260224506201e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8659312095227427e-08, + "pre_projection_traffic_rms_ratio": 0.004901452516407269 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.955478668212891, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.6436767578125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07829976826906204, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.783659815788269, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1818346083164215, + "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": 0.6048848628997803, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1309211105108261, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.782848834991455, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13312062621116638, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11671629548072815, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7305636405944824, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18173032999038696, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1667901277542114, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019707441514618004, + "step": 135, + "teaching_rms": 0.0017763656391565899, + "traffic_rms": 0.01962846582448694 + }, + { + "batch_loss": 1.7838380336761475, + "innovation_rms": 0.0015173786362940097, + "instruction_rms": 0.0015173786395515393, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 1.0073820000400246e-08, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 1.0073820000400246e-08, + "min_post_projection_soma_slope": -4.937104058910791e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8580342261899767e-08, + "pre_projection_traffic_rms_ratio": 0.004986278833388844 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 6.98997163772583, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.12689971923828, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0674065351486206, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7865535020828247, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21243315935134888, + "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": 0.6101015210151672, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10602466762065887, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7831106185913086, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07578865438699722, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09670208394527435, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7259909510612488, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21232889592647552, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1667354106903076, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019747796756468573, + "step": 136, + "teaching_rms": 0.0015173786362940097, + "traffic_rms": 0.01968860197721461 + }, + { + "batch_loss": 1.9542605876922607, + "innovation_rms": 0.001724653765719418, + "instruction_rms": 0.0017246537671524321, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 7.539592061789335e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.9298480853110505e-09, + "min_post_projection_soma_slope": -7.539592061789335e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8772956736203558e-08, + "pre_projection_traffic_rms_ratio": 0.005079209441696844 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.030772686004639, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.7314224243164, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.061890773475170135, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7824700474739075, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18771511316299438, + "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": 0.6110512614250183, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12863746285438538, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7800463438034058, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10598024725914001, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10419657826423645, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7288587093353271, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18758264183998108, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166725516319275, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019745785363185865, + "step": 137, + "teaching_rms": 0.001724653765719418, + "traffic_rms": 0.019671854630489656 + }, + { + "batch_loss": 1.7564493417739868, + "innovation_rms": 0.0015170622434240203, + "instruction_rms": 0.0015170622450531247, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 7.182709538255949e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 7.182709538255949e-09, + "min_post_projection_soma_slope": -5.57481838470153e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.862789246194376e-08, + "pre_projection_traffic_rms_ratio": 0.004844673538828865 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.064223289489746, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.36017608642578, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07156242430210114, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7783596515655518, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20307141542434692, + "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": 0.6151237487792969, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1326611489057541, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7769697904586792, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11024987697601318, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09469720721244812, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7335562109947205, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2029389590024948, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1666486263275146, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019717447691562123, + "step": 138, + "teaching_rms": 0.0015170622434240203, + "traffic_rms": 0.01966024301862125 + }, + { + "batch_loss": 1.8211288452148438, + "innovation_rms": 0.0016421459582732908, + "instruction_rms": 0.0016421459582732908, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 6.812880037898594e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 6.812880037898594e-09, + "min_post_projection_soma_slope": -6.166093857729038e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.868114889517216e-08, + "pre_projection_traffic_rms_ratio": 0.005044576105375117 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.100567817687988, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.12682342529297, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05693693459033966, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7747188806533813, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1528797596693039, + "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": 0.6230102181434631, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11011504381895065, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7808756828308105, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13363094627857208, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09115888923406601, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.740273118019104, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15274731814861298, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1665594577789307, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019729628427915303, + "step": 139, + "teaching_rms": 0.0016421459582732908, + "traffic_rms": 0.019658906613990293 + }, + { + "batch_loss": 1.7631276845932007, + "innovation_rms": 0.0015435379216195248, + "instruction_rms": 0.0015435379168160413, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.483457687915916e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.483457687915916e-09, + "min_post_projection_soma_slope": -4.382597840191238e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8692112276925735e-08, + "pre_projection_traffic_rms_ratio": 0.004946989567351498 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.128930568695068, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.77482604980469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07193110883235931, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7697266936302185, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14603030681610107, + "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": 0.6335062384605408, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10496008396148682, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7893056869506836, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17061327397823334, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12982052564620972, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7394410967826843, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14624282717704773, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1662768125534058, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01975883825773756, + "step": 140, + "teaching_rms": 0.0015435379216195248, + "traffic_rms": 0.01969708995468076 + }, + { + "batch_loss": 1.7647454738616943, + "innovation_rms": 0.0013621635044943068, + "instruction_rms": 0.0013621634972368726, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 1.2473076793639848e-08, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.2695904589606926e-09, + "min_post_projection_soma_slope": -1.2473076793639848e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8660132415595285e-08, + "pre_projection_traffic_rms_ratio": 0.0045876160548466854 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.157533645629883, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.46890258789062, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06003626808524132, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7656612992286682, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16651666164398193, + "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": 0.6473655700683594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13859355449676514, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7954657077789307, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10902251303195953, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11710894852876663, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7353725433349609, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1667291671037674, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1659609079360962, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019721392263606273, + "step": 141, + "teaching_rms": 0.0013621635044943068, + "traffic_rms": 0.019673792624293616 + }, + { + "batch_loss": 1.8029069900512695, + "innovation_rms": 0.0013733427331983172, + "instruction_rms": 0.001373342736797496, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967561036348343, + "max_absolute_post_projection_soma_slope": 1.0427642749277766e-08, + "max_absolute_pre_projection_soma_slope": 0.02967561036348343, + "max_positive_post_projection_soma_slope": 1.0427642749277766e-08, + "min_post_projection_soma_slope": -3.9867447121366695e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8564287184269863e-08, + "pre_projection_traffic_rms_ratio": 0.004844029765555727 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.185079097747803, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.98900604248047, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05632542073726654, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.764133095741272, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22474196553230286, + "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": 0.6643236875534058, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1695813536643982, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8023513555526733, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10033664852380753, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09953150153160095, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7365803122520447, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22506001591682434, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1657140254974365, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019773895085352006, + "step": 142, + "teaching_rms": 0.0013733427331983172, + "traffic_rms": 0.019723903139291413 + }, + { + "batch_loss": 1.7341837882995605, + "innovation_rms": 0.0014216143208341475, + "instruction_rms": 0.0014216143190956638, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.79840389644437e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.79840389644437e-09, + "min_post_projection_soma_slope": -4.572122680457369e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8729334246708308e-08, + "pre_projection_traffic_rms_ratio": 0.0048146629098359305 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.213216304779053, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.2968521118164, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05015869811177254, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.76360023021698, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14740273356437683, + "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": 0.6779543161392212, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13630598783493042, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8072243928909302, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09097201377153397, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09543586522340775, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7390115857124329, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14772075414657593, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1658146381378174, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019759545371361303, + "step": 143, + "teaching_rms": 0.0014216143208341475, + "traffic_rms": 0.019707850703689002 + }, + { + "batch_loss": 1.8602030277252197, + "innovation_rms": 0.0014974181381374766, + "instruction_rms": 0.0014974181381374766, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.134213054702741e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 2.603814275303762e-09, + "min_post_projection_soma_slope": -5.134213054702741e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8773019725061184e-08, + "pre_projection_traffic_rms_ratio": 0.004806484937803521 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.249092102050781, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.5196304321289, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.048017702996730804, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7637612223625183, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14978478848934174, + "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": 0.6905325055122375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12578196823596954, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8095653057098389, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0864819660782814, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09723053127527237, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.740643322467804, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14981985092163086, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165574550628662, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019773618372201224, + "step": 144, + "teaching_rms": 0.0014974181381374766, + "traffic_rms": 0.019717242839685928 + }, + { + "batch_loss": 1.8698766231536865, + "innovation_rms": 0.0015847078563256377, + "instruction_rms": 0.0015847078500873825, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 7.077540775668467e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.6315914719864395e-09, + "min_post_projection_soma_slope": -7.077540775668467e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.869089204955719e-08, + "pre_projection_traffic_rms_ratio": 0.004962985475484332 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.283244609832764, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.41944885253906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08221343904733658, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7638936638832092, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14289957284927368, + "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": 0.696621835231781, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06825199723243713, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.810085654258728, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08331983536481857, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0820290744304657, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7393890619277954, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14307262003421783, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1653153896331787, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019763370818268292, + "step": 145, + "teaching_rms": 0.0015847078563256377, + "traffic_rms": 0.019701235211705304 + }, + { + "batch_loss": 1.754738688468933, + "innovation_rms": 0.0015974922561417613, + "instruction_rms": 0.0015974922515005127, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 3.082833988088396e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.082833988088396e-09, + "min_post_projection_soma_slope": -3.0820412888488136e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8667853107864728e-08, + "pre_projection_traffic_rms_ratio": 0.005065353424999744 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.310457229614258, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.41071319580078, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06529224663972855, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7646685838699341, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1261817365884781, + "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": 0.7043527960777283, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08982664346694946, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.811968207359314, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09078707545995712, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0806926041841507, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7358800768852234, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12649965286254883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1652265787124634, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019764537691915315, + "step": 146, + "teaching_rms": 0.0015974922561417613, + "traffic_rms": 0.019698493640391807 + }, + { + "batch_loss": 1.7018742561340332, + "innovation_rms": 0.0016207987016141432, + "instruction_rms": 0.0016207986894154513, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 8.018279373800397e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 8.018279373800397e-09, + "min_post_projection_soma_slope": -2.622518424644227e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.868861130590223e-08, + "pre_projection_traffic_rms_ratio": 0.005149757648036297 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.346322059631348, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.67376708984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07782412320375443, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7662943005561829, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1398146152496338, + "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": 0.7066329121589661, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1071045994758606, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8099764585494995, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08244473487138748, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08024764806032181, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7355069518089294, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14012175798416138, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1654301881790161, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019754721060596517, + "step": 147, + "teaching_rms": 0.0016207987016141432, + "traffic_rms": 0.0196883325610881 + }, + { + "batch_loss": 1.8867372274398804, + "innovation_rms": 0.0017521744946972196, + "instruction_rms": 0.0017521745059812693, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.311102224681008e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.04586009139507e-09, + "min_post_projection_soma_slope": -5.311102224681008e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.882039266725381e-08, + "pre_projection_traffic_rms_ratio": 0.004753105873292113 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.3822197914123535, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.20689392089844, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0672646015882492, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7671502232551575, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17438484728336334, + "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": 0.6997565627098083, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12636588513851166, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8047641515731812, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12890835106372833, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14102774858474731, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7299827337265015, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17406697571277618, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165833830833435, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019766205736388938, + "step": 148, + "teaching_rms": 0.0017521744946972196, + "traffic_rms": 0.01969163780774105 + }, + { + "batch_loss": 1.7545489072799683, + "innovation_rms": 0.0018752980332578972, + "instruction_rms": 0.0018752980227147067, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.6230383138047273e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.6230383138047273e-09, + "min_post_projection_soma_slope": -2.201259619027951e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8643984658706418e-08, + "pre_projection_traffic_rms_ratio": 0.005172898230018377 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.412553310394287, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.68202209472656, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07878722250461578, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.769656240940094, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19178453087806702, + "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": 0.6929805874824524, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12193737924098969, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7939748764038086, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10789300501346588, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10665683448314667, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7281099557876587, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.191822350025177, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1661535501480103, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019814125054585532, + "step": 149, + "teaching_rms": 0.0018752980332578972, + "traffic_rms": 0.01972569186164144 + }, + { + "batch_loss": 1.7991697788238525, + "innovation_rms": 0.0018784048639542727, + "instruction_rms": 0.0018784048718485868, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 3.755189048604279e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.755189048604279e-09, + "min_post_projection_soma_slope": -3.2527678328619913e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.877195988696082e-08, + "pre_projection_traffic_rms_ratio": 0.004784385557503769 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.437694072723389, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.11295318603516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10501042008399963, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7714297771453857, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16657668352127075, + "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": 0.6855002045631409, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13571712374687195, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7814935445785522, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12481290847063065, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1051824614405632, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.735224187374115, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16625887155532837, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1664992570877075, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019842060491475237, + "step": 150, + "teaching_rms": 0.0018784048639542727, + "traffic_rms": 0.019750141125727944 + }, + { + "batch_loss": 1.7606984376907349, + "innovation_rms": 0.001669293766787271, + "instruction_rms": 0.001669293768267809, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.263820046508272e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.263820046508272e-09, + "min_post_projection_soma_slope": -3.9341694346717304e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.879389751914647e-08, + "pre_projection_traffic_rms_ratio": 0.00468805193048912 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.4631171226501465, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.26031494140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11240766942501068, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7733873128890991, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1653308868408203, + "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": 0.6748456954956055, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15591780841350555, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7690926790237427, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12400840222835541, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11257969588041306, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7389937043190002, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16514670848846436, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1669609546661377, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019856046592138587, + "step": 151, + "teaching_rms": 0.001669293766787271, + "traffic_rms": 0.019786891804795346 + }, + { + "batch_loss": 1.8243345022201538, + "innovation_rms": 0.0015940826308518696, + "instruction_rms": 0.0015940826230999095, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 7.135336321795194e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 7.135336321795194e-09, + "min_post_projection_soma_slope": -4.717017887401198e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.869633175204094e-08, + "pre_projection_traffic_rms_ratio": 0.004914071841114105 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.488102912902832, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.7439956665039, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09868587553501129, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7804929614067078, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13641642034053802, + "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": 0.6702162027359009, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14349985122680664, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7614281177520752, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11354250460863113, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09885788708925247, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7359640002250671, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13661691546440125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1674267053604126, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.019938230787290404, + "step": 152, + "teaching_rms": 0.0015940826308518696, + "traffic_rms": 0.01987147170456006 + }, + { + "batch_loss": 1.8505663871765137, + "innovation_rms": 0.0016237295808928898, + "instruction_rms": 0.0016237295778487213, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.062228775580934e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.733677367279142e-09, + "min_post_projection_soma_slope": -4.062228775580934e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8559791795628356e-08, + "pre_projection_traffic_rms_ratio": 0.004942181265137721 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.501972198486328, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.5931167602539, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07806423306465149, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.788952112197876, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24393624067306519, + "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": 0.667434811592102, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12360215187072754, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7579385042190552, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10790427774190903, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1008564904332161, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7218214273452759, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2440403252840042, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.168149709701538, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020028494413360867, + "step": 153, + "teaching_rms": 0.0016237295808928898, + "traffic_rms": 0.019963882602250706 + }, + { + "batch_loss": 1.8581385612487793, + "innovation_rms": 0.001633945839119343, + "instruction_rms": 0.0016339458436570445, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 2.861123338249172e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.673638421768487e-09, + "min_post_projection_soma_slope": -2.861123338249172e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8527322694963473e-08, + "pre_projection_traffic_rms_ratio": 0.00472709076290612 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.515554428100586, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.9865951538086, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0668865442276001, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7981409430503845, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23800531029701233, + "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": 0.667069673538208, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11751340329647064, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7598888874053955, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10033483803272247, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0830608606338501, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6993500590324402, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23810938000679016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.168848991394043, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020102989860024244, + "step": 154, + "teaching_rms": 0.001633945839119343, + "traffic_rms": 0.020037650324929147 + }, + { + "batch_loss": 1.8430585861206055, + "innovation_rms": 0.0017491397395073206, + "instruction_rms": 0.0017491397395073206, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.900965411280822e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.900965411280822e-09, + "min_post_projection_soma_slope": -4.874558534595508e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8655392265180195e-08, + "pre_projection_traffic_rms_ratio": 0.004594483751113003 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.538992404937744, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.76827239990234, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05172284319996834, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8065804243087769, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20979925990104675, + "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": 0.6697408556938171, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09957842528820038, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.760766863822937, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06628520041704178, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0628669410943985, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6845767498016357, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20973823964595795, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1694409847259521, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02005747011051082, + "step": 155, + "teaching_rms": 0.0017491397395073206, + "traffic_rms": 0.019980721620966767 + }, + { + "batch_loss": 1.8367159366607666, + "innovation_rms": 0.0018067694402676732, + "instruction_rms": 0.0018067694347961329, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 5.9031619592531115e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.8941637409427585e-09, + "min_post_projection_soma_slope": -5.9031619592531115e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.861117718542922e-08, + "pre_projection_traffic_rms_ratio": 0.004708856714047735 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.563884258270264, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.0189437866211, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.053915224969387054, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8120414018630981, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1617830991744995, + "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": 0.6700339913368225, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1131589338183403, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7613428831100464, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06918837130069733, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0573381707072258, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6831698417663574, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1617220789194107, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1697325706481934, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02013920708038524, + "step": 156, + "teaching_rms": 0.0018067694402676732, + "traffic_rms": 0.020057011996340593 + }, + { + "batch_loss": 1.7482162714004517, + "innovation_rms": 0.001618843682454837, + "instruction_rms": 0.001618843676348125, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.896476196198819e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 5.896476196198819e-09, + "min_post_projection_soma_slope": -3.341674714718579e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8578541342854724e-08, + "pre_projection_traffic_rms_ratio": 0.004730477791629106 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.583195209503174, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.76717376708984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05050084367394447, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8152696490287781, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21411971747875214, + "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": 0.6725854277610779, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07877478003501892, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7635129690170288, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08606431633234024, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05023222416639328, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6916568279266357, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21405869722366333, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1696456670761108, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02007730155576793, + "step": 157, + "teaching_rms": 0.001618843682454837, + "traffic_rms": 0.020011012026240886 + }, + { + "batch_loss": 1.693407416343689, + "innovation_rms": 0.0015862167598438884, + "instruction_rms": 0.0015862167723085304, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.190581298109009e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.8290625397886515e-09, + "min_post_projection_soma_slope": -5.190581298109009e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8772121225891547e-08, + "pre_projection_traffic_rms_ratio": 0.004404104670959384 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.60398530960083, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.83392333984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047536034137010574, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8154142498970032, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19626091420650482, + "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": 0.6759078502655029, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08115344494581223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7650383710861206, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10565072298049927, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04726744070649147, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7099741101264954, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19622313976287842, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1690891981124878, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020104399239063205, + "step": 158, + "teaching_rms": 0.0015862167598438884, + "traffic_rms": 0.020039897476892974 + }, + { + "batch_loss": 1.7349331378936768, + "innovation_rms": 0.0015864480991932813, + "instruction_rms": 0.0015864481038668403, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.188192903598065e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 1.6392802670850415e-09, + "min_post_projection_soma_slope": -4.188192903598065e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.857204228700403e-08, + "pre_projection_traffic_rms_ratio": 0.00477374916517546 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.622306823730469, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.76689910888672, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.053727492690086365, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8147168159484863, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17444467544555664, + "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": 0.6800212264060974, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07058975100517273, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.765714168548584, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09239310771226883, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07585716247558594, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7274100184440613, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17435921728610992, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1686732769012451, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02008512913858307, + "step": 159, + "teaching_rms": 0.0015864480991932813, + "traffic_rms": 0.02002022127110207 + }, + { + "batch_loss": 1.8773367404937744, + "innovation_rms": 0.0016678738628338806, + "instruction_rms": 0.0016678738539430897, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.0713884768829303e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.0713884768829303e-09, + "min_post_projection_soma_slope": -2.678472332817705e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8680837650955575e-08, + "pre_projection_traffic_rms_ratio": 0.004600473222618053 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6404128074646, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.54730224609375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06649221479892731, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8143803477287292, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19289150834083557, + "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": 0.6869444847106934, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06923229247331619, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7673894166946411, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08958286792039871, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08369510620832443, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.736426055431366, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19277790188789368, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.168310284614563, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020118309567659778, + "step": 160, + "teaching_rms": 0.0016678738628338806, + "traffic_rms": 0.020048879893558115 + }, + { + "batch_loss": 1.7241625785827637, + "innovation_rms": 0.0016434542681915312, + "instruction_rms": 0.0016434542711991633, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.897327515214101e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.9405311519070665e-09, + "min_post_projection_soma_slope": -5.897327515214101e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8559223255830002e-08, + "pre_projection_traffic_rms_ratio": 0.004579134592073092 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.655964374542236, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.45526885986328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07341068238019943, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8118177056312561, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24620601534843445, + "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": 0.6955927014350891, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08648240566253662, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7696954011917114, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06888533383607864, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08679454028606415, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7327117323875427, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24626702070236206, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1678944826126099, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02013273152000182, + "step": 161, + "teaching_rms": 0.0016434542681915312, + "traffic_rms": 0.020065483527962214 + }, + { + "batch_loss": 1.8737080097198486, + "innovation_rms": 0.0017980063940814795, + "instruction_rms": 0.0017980064023287905, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.748525211402011e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 5.748525211402011e-09, + "min_post_projection_soma_slope": -2.358185646755828e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.853184501525033e-08, + "pre_projection_traffic_rms_ratio": 0.004703547613765039 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.673534870147705, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.76985168457031, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0629432275891304, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8087538480758667, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16004562377929688, + "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": 0.7010965943336487, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07128868997097015, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7692288160324097, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09378310292959213, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08135560154914856, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7209892868995667, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1601066291332245, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.167559266090393, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020174652255431344, + "step": 162, + "teaching_rms": 0.0017980063940814795, + "traffic_rms": 0.02009736024111679 + }, + { + "batch_loss": 1.7810746431350708, + "innovation_rms": 0.0015974233352048374, + "instruction_rms": 0.0015974233429405858, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 5.452188034382743e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 5.452188034382743e-09, + "min_post_projection_soma_slope": -1.9978254606201062e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8650386056013974e-08, + "pre_projection_traffic_rms_ratio": 0.004403216011935916 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.691014289855957, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.37828063964844, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05098820850253105, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8066441416740417, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14387749135494232, + "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": 0.7065142393112183, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09485883265733719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7710140943527222, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09856770932674408, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07393565773963928, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7077463865280151, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1440877616405487, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.16734778881073, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020134536254096336, + "step": 163, + "teaching_rms": 0.0015974233352048374, + "traffic_rms": 0.020074822054425796 + }, + { + "batch_loss": 1.7952322959899902, + "innovation_rms": 0.0017816083114556983, + "instruction_rms": 0.0017816083086812915, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 6.9323951024102826e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 6.9323951024102826e-09, + "min_post_projection_soma_slope": -4.333926550970091e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8539784417493467e-08, + "pre_projection_traffic_rms_ratio": 0.005035888080735394 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.706931114196777, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.43041229248047, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04340558126568794, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8082405924797058, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15053978562355042, + "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": 0.705734372138977, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08230599761009216, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7717357873916626, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06824222952127457, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06059730798006058, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6994965076446533, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15050479769706726, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1671926975250244, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02017825900581647, + "step": 164, + "teaching_rms": 0.0017816083114556983, + "traffic_rms": 0.020101646102297817 + }, + { + "batch_loss": 1.8923145532608032, + "innovation_rms": 0.001834553343289865, + "instruction_rms": 0.0018345533446370336, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.05718186719173e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.021321498015595e-09, + "min_post_projection_soma_slope": -6.05718186719173e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8492240230059507e-08, + "pre_projection_traffic_rms_ratio": 0.004813342067298457 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.718258857727051, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.36637115478516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0496983528137207, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.810063362121582, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16477523744106293, + "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": 0.7046980261802673, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09433393180370331, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.771892786026001, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08438289165496826, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05672840774059296, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6921648979187012, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16445791721343994, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166981816291809, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02013740067358436, + "step": 165, + "teaching_rms": 0.001834553343289865, + "traffic_rms": 0.02005756366523201 + }, + { + "batch_loss": 1.7250405550003052, + "innovation_rms": 0.0017299758416757017, + "instruction_rms": 0.0017299758373898852, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.3442112957213794e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.606180243309609e-09, + "min_post_projection_soma_slope": -5.3442112957213794e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8536223905569985e-08, + "pre_projection_traffic_rms_ratio": 0.004685161027592924 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.716619491577148, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.48372650146484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05136178433895111, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8112595081329346, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1753235161304474, + "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": 0.7006840705871582, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0872192531824112, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7666099071502686, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.080594003200531, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05948773771524429, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6937270164489746, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17526254057884216, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1666754484176636, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02012639182632837, + "step": 166, + "teaching_rms": 0.0017299758416757017, + "traffic_rms": 0.0200521617903931 + }, + { + "batch_loss": 1.679337501525879, + "innovation_rms": 0.001796223813596335, + "instruction_rms": 0.0017962238190999988, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.496873679509349e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.496873679509349e-09, + "min_post_projection_soma_slope": -2.8357289849623157e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8601269554315714e-08, + "pre_projection_traffic_rms_ratio": 0.004807265067285681 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.718822479248047, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.47758483886719, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04782847315073013, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8115097284317017, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18294887244701385, + "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": 0.6964089870452881, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10701330006122589, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7625627517700195, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0799219161272049, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05489011108875275, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7005751132965088, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18288789689540863, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166359305381775, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0201003564310506, + "step": 167, + "teaching_rms": 0.001796223813596335, + "traffic_rms": 0.020022706475347103 + }, + { + "batch_loss": 1.6962766647338867, + "innovation_rms": 0.0018612267427552565, + "instruction_rms": 0.0018612267427552565, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 2.488791173149707e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 2.2382640185725222e-09, + "min_post_projection_soma_slope": -2.488791173149707e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8500192893222245e-08, + "pre_projection_traffic_rms_ratio": 0.004816243704138403 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.717648983001709, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.81590270996094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0736142247915268, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8115221261978149, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14549510180950165, + "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": 0.687169075012207, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1058417409658432, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7541447877883911, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08821302652359009, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08214525878429413, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7110873460769653, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14574620127677917, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1658998727798462, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020146233954926363, + "step": 168, + "teaching_rms": 0.0018612267427552565, + "traffic_rms": 0.020059336641236106 + }, + { + "batch_loss": 1.886103868484497, + "innovation_rms": 0.0018047724965893628, + "instruction_rms": 0.0018047724965893628, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 1.3425393241561778e-08, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 1.3425393241561778e-08, + "min_post_projection_soma_slope": -6.910793270975546e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8499759277219884e-08, + "pre_projection_traffic_rms_ratio": 0.004855014987767544 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.715792655944824, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.91712951660156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06512177735567093, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8111028075218201, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19536641240119934, + "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": 0.6798694133758545, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09339871257543564, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.747405767440796, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08090579509735107, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08169181644916534, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7306154370307922, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19528105854988098, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1654791831970215, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020095094601578607, + "step": 169, + "teaching_rms": 0.0018047724965893628, + "traffic_rms": 0.02001152194128811 + }, + { + "batch_loss": 1.8258777856826782, + "innovation_rms": 0.0017115707094492005, + "instruction_rms": 0.0017115707152250718, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 6.474405012113493e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.3659632858729083e-09, + "min_post_projection_soma_slope": -6.474405012113493e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8556335535209552e-08, + "pre_projection_traffic_rms_ratio": 0.004723910657058485 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.713578701019287, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.02405548095703, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06283896416425705, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8102933764457703, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18908831477165222, + "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": 0.676068127155304, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.100880466401577, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7409714460372925, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06820434331893921, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07067267596721649, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7495157122612, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18900297582149506, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1650102138519287, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020110805406333233, + "step": 170, + "teaching_rms": 0.0017115707094492005, + "traffic_rms": 0.020036960039638917 + }, + { + "batch_loss": 1.6692906618118286, + "innovation_rms": 0.0017189095399113951, + "instruction_rms": 0.0017189095514138178, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.307388667967871e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.307388667967871e-09, + "min_post_projection_soma_slope": -4.243217333055327e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8581472973933982e-08, + "pre_projection_traffic_rms_ratio": 0.004516647690427012 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.706928253173828, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.22135162353516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06121433153748512, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8109816908836365, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17365474998950958, + "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": 0.6776109933853149, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1042604073882103, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7406928539276123, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07164446264505386, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07443338632583618, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7668726444244385, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1735694259405136, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1646963357925415, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020066438108776903, + "step": 171, + "teaching_rms": 0.0017189095399113951, + "traffic_rms": 0.01999036075080392 + }, + { + "batch_loss": 1.8035218715667725, + "innovation_rms": 0.0019604464815078894, + "instruction_rms": 0.001960446478986573, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 5.9225024884312916e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 5.9225024884312916e-09, + "min_post_projection_soma_slope": -2.0197901129392903e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8512793906541027e-08, + "pre_projection_traffic_rms_ratio": 0.004696029755794483 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.707092761993408, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.3208999633789, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07429707050323486, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8087047338485718, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14392589032649994, + "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": 0.6789310574531555, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12067317962646484, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7366551160812378, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07021814584732056, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07408319413661957, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7678069472312927, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14424295723438263, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1643006801605225, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02005740863321671, + "step": 172, + "teaching_rms": 0.0019604464815078894, + "traffic_rms": 0.019960581609440763 + }, + { + "batch_loss": 1.6665992736816406, + "innovation_rms": 0.001695186453119251, + "instruction_rms": 0.001695186460408871, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 4.410198428672629e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.410198428672629e-09, + "min_post_projection_soma_slope": -2.2686965639451273e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.844160291347282e-08, + "pre_projection_traffic_rms_ratio": 0.004426596583069613 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.697033882141113, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.46595001220703, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07634344696998596, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8048181533813477, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16737499833106995, + "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": 0.6813822388648987, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12474856525659561, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7323676347732544, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07352832704782486, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07612960040569305, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7514203786849976, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1674099564552307, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1641254425048828, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020035390691084957, + "step": 173, + "teaching_rms": 0.001695186453119251, + "traffic_rms": 0.019963717258853354 + }, + { + "batch_loss": 1.930551528930664, + "innovation_rms": 0.0019205746551690013, + "instruction_rms": 0.0019205746551690013, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 8.344528623638325e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.039947487655127e-09, + "min_post_projection_soma_slope": -8.344528623638325e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.854000146049604e-08, + "pre_projection_traffic_rms_ratio": 0.004463463820093465 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.689632415771484, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.7239761352539, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06427311152219772, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7975616455078125, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21684718132019043, + "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": 0.680009126663208, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1377214789390564, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7342345714569092, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1012781411409378, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06738758832216263, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7320439219474792, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21685022115707397, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1640074253082275, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020102166620115297, + "step": 174, + "teaching_rms": 0.0019205746551690013, + "traffic_rms": 0.020009670455012372 + }, + { + "batch_loss": 1.8639811277389526, + "innovation_rms": 0.0019590647783639618, + "instruction_rms": 0.0019590647783639618, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.96293273144488e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.96293273144488e-09, + "min_post_projection_soma_slope": -1.9857528954503323e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8591359495256716e-08, + "pre_projection_traffic_rms_ratio": 0.004481459239706456 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6826558113098145, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.99072265625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10507500171661377, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7927170991897583, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1723523586988449, + "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": 0.6902809143066406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14399096369743347, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7292649745941162, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1108151227235794, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10486119985580444, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7168018221855164, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17235538363456726, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1639158725738525, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0200494014952973, + "step": 175, + "teaching_rms": 0.0019590647783639618, + "traffic_rms": 0.01995451114058166 + }, + { + "batch_loss": 1.8323042392730713, + "innovation_rms": 0.0018824262227787958, + "instruction_rms": 0.0018824262201529794, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 6.133084706760883e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 6.133084706760883e-09, + "min_post_projection_soma_slope": -5.482597487116436e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8485426450916004e-08, + "pre_projection_traffic_rms_ratio": 0.004591535139465058 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.664393424987793, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.18168640136719, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09855980426073074, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7898189425468445, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12623752653598785, + "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": 0.7028226256370544, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14387162029743195, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7274887561798096, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10017095506191254, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09834601730108261, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7068113088607788, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.126554474234581, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1639012098312378, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020095987999099726, + "step": 176, + "teaching_rms": 0.0018824262227787958, + "traffic_rms": 0.02000884480246886 + }, + { + "batch_loss": 1.991847276687622, + "innovation_rms": 0.002097606491627564, + "instruction_rms": 0.0020976065151920693, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 9.412733703584308e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 9.412733703584308e-09, + "min_post_projection_soma_slope": -3.1501301567260498e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.864354831022578e-08, + "pre_projection_traffic_rms_ratio": 0.00463177431056815 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.645825386047363, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.42809295654297, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10810413211584091, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7883062362670898, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13974003493785858, + "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": 0.7154318690299988, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1260925680398941, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7243435382843018, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12732359766960144, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10812383890151978, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7057509422302246, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13981589674949646, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1638166904449463, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020092481361399947, + "step": 177, + "teaching_rms": 0.002097606491627564, + "traffic_rms": 0.01998947940859749 + }, + { + "batch_loss": 1.7948652505874634, + "innovation_rms": 0.0018997771039662085, + "instruction_rms": 0.001899777098762539, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 1.027989515733907e-08, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 6.018324505419059e-09, + "min_post_projection_soma_slope": -1.027989515733907e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8545183838237622e-08, + "pre_projection_traffic_rms_ratio": 0.004866683104617201 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.631372451782227, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.31660461425781, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08266683667898178, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7878684997558594, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14761137962341309, + "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": 0.7234352827072144, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08003436774015427, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7254923582077026, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0987185537815094, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08375972509384155, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7086694836616516, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14753663539886475, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1641136407852173, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020099328506101747, + "step": 178, + "teaching_rms": 0.0018997771039662085, + "traffic_rms": 0.020007610969554553 + }, + { + "batch_loss": 1.8706094026565552, + "innovation_rms": 0.002179435547947626, + "instruction_rms": 0.0021794355502156013, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.67720528973814e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.4504255008727114e-09, + "min_post_projection_soma_slope": -4.67720528973814e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8669041677484165e-08, + "pre_projection_traffic_rms_ratio": 0.0049856060598266775 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.621018409729004, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.6347427368164, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08920052647590637, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7890428900718689, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15579716861248016, + "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": 0.7251238822937012, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10388346016407013, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.737355351448059, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12910932302474976, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1286991983652115, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.718400776386261, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15600642561912537, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1646897792816162, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02009540464178208, + "step": 179, + "teaching_rms": 0.002179435547947626, + "traffic_rms": 0.019978309825163524 + }, + { + "batch_loss": 1.7301274538040161, + "innovation_rms": 0.0020558483151183913, + "instruction_rms": 0.002055848310309762, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 7.232355603292717e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 7.232355603292717e-09, + "min_post_projection_soma_slope": -3.0218987312480294e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.849452554102716e-08, + "pre_projection_traffic_rms_ratio": 0.004978127420106556 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.614166736602783, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.59103393554688, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09815193712711334, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7908366918563843, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17432306706905365, + "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": 0.7302972078323364, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11502356827259064, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7473220825195312, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11276604235172272, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16399095952510834, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7256931066513062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17415064573287964, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1651290655136108, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020075091730106704, + "step": 180, + "teaching_rms": 0.0020558483151183913, + "traffic_rms": 0.019970406604204766 + }, + { + "batch_loss": 1.6659584045410156, + "innovation_rms": 0.0018383501667712022, + "instruction_rms": 0.0018383501640824294, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.655125174224395e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.831962747113948e-09, + "min_post_projection_soma_slope": -4.655125174224395e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.863606166744318e-08, + "pre_projection_traffic_rms_ratio": 0.005015954929201822 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.604523181915283, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.56391906738281, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08252564817667007, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7892841100692749, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15178053081035614, + "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": 0.7362875938415527, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09688510000705719, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7515487670898438, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.085425466299057, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10601168125867844, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7277165651321411, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15146373212337494, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1653382778167725, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01997079145865693, + "step": 181, + "teaching_rms": 0.0018383501667712022, + "traffic_rms": 0.019884391969498615 + }, + { + "batch_loss": 1.8106861114501953, + "innovation_rms": 0.002176923247171909, + "instruction_rms": 0.0021769232517130943, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 9.728298167033245e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 9.728298167033245e-09, + "min_post_projection_soma_slope": -2.8912703342598434e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.855729548634874e-08, + "pre_projection_traffic_rms_ratio": 0.004943472170296783 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.592439651489258, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.3038330078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0739271342754364, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7868331670761108, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12116383016109467, + "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": 0.7408937811851501, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08641980588436127, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7514275312423706, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07119802385568619, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.075006403028965, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7252354621887207, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12099145352840424, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165416955947876, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020109429788156018, + "step": 182, + "teaching_rms": 0.002176923247171909, + "traffic_rms": 0.019989492080961818 + }, + { + "batch_loss": 1.7284621000289917, + "innovation_rms": 0.002028784899038444, + "instruction_rms": 0.0020287848844201192, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 4.427744837443015e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.427744837443015e-09, + "min_post_projection_soma_slope": -3.7957006426836415e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8618051041957855e-08, + "pre_projection_traffic_rms_ratio": 0.005036108520114321 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.596353054046631, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.81980895996094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06412280350923538, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7854791879653931, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10606725513935089, + "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": 0.7453771829605103, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07451116293668747, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7483000755310059, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13694258034229279, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09276536107063293, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7270677089691162, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10593539476394653, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1651384830474854, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01994718828627106, + "step": 183, + "teaching_rms": 0.002028784899038444, + "traffic_rms": 0.019842396249484237 + }, + { + "batch_loss": 1.6672686338424683, + "innovation_rms": 0.0020128943144195654, + "instruction_rms": 0.0020128943144195654, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 5.460894403341854e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 3.892976607744458e-09, + "min_post_projection_soma_slope": -5.460894403341854e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8706346497383577e-08, + "pre_projection_traffic_rms_ratio": 0.005374021048679763 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.598387718200684, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.4518814086914, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05861051380634308, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7840830087661743, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16914790868759155, + "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": 0.7441865801811218, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09320443868637085, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7520040273666382, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10601881891489029, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08822476863861084, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7306947112083435, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1689387559890747, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1647855043411255, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.01994683975997722, + "step": 184, + "teaching_rms": 0.0020128943144195654, + "traffic_rms": 0.019840819049147147 + }, + { + "batch_loss": 1.828383445739746, + "innovation_rms": 0.002083149998730592, + "instruction_rms": 0.0020831500177130213, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.897495215454683e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.897495215454683e-09, + "min_post_projection_soma_slope": -3.208355581207911e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8728405221619716e-08, + "pre_projection_traffic_rms_ratio": 0.005136877566670812 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.593266010284424, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.14685821533203, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05314397066831589, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7831633687019348, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12960314750671387, + "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": 0.7462021112442017, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09850087016820908, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7472277879714966, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.130989670753479, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09936937689781189, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7260522842407227, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1293940246105194, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1645712852478027, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02001566770176861, + "step": 185, + "teaching_rms": 0.002083149998730592, + "traffic_rms": 0.019908605720235485 + }, + { + "batch_loss": 1.5849450826644897, + "innovation_rms": 0.0019112878579852755, + "instruction_rms": 0.0019112878541060279, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 5.340291320266033e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 5.026060900803486e-09, + "min_post_projection_soma_slope": -5.340291320266033e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8562809085866593e-08, + "pre_projection_traffic_rms_ratio": 0.005043583978175326 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.589836120605469, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.7193603515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05958038568496704, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7822549343109131, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1684597283601761, + "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": 0.7509273290634155, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08670267462730408, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7472715377807617, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12811022996902466, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13607053458690643, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7221882343292236, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16848433017730713, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.164487600326538, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02004307826909513, + "step": 186, + "teaching_rms": 0.0019112878579852755, + "traffic_rms": 0.019948503463764405 + }, + { + "batch_loss": 1.5806478261947632, + "innovation_rms": 0.0019467164697432656, + "instruction_rms": 0.001946716454508671, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 9.957332736121316e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.311264234502232e-09, + "min_post_projection_soma_slope": -9.957332736121316e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.859559565106009e-08, + "pre_projection_traffic_rms_ratio": 0.005054993125881503 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.590625762939453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.67195892333984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08872620761394501, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7823286056518555, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21458187699317932, + "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": 0.7532586455345154, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0856148973107338, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.750938892364502, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10059607028961182, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11941248923540115, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7165069580078125, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21460649371147156, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1643781661987305, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020023166937513978, + "step": 187, + "teaching_rms": 0.0019467164697432656, + "traffic_rms": 0.01992739764810921 + }, + { + "batch_loss": 1.7014217376708984, + "innovation_rms": 0.002288660896195655, + "instruction_rms": 0.002288660902674867, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 9.107853138345945e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 5.097305244561312e-09, + "min_post_projection_soma_slope": -9.107853138345945e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.876501689369416e-08, + "pre_projection_traffic_rms_ratio": 0.00512654106776334 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.596382141113281, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.77497863769531, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09784874320030212, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.781569242477417, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15927140414714813, + "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": 0.7531715631484985, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08378073573112488, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7566486597061157, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07736940681934357, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11430570483207703, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7121044993400574, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15933765470981598, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.164415955543518, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020052418324784854, + "step": 188, + "teaching_rms": 0.002288660896195655, + "traffic_rms": 0.01991663835531878 + }, + { + "batch_loss": 1.694116473197937, + "innovation_rms": 0.0021735277500717527, + "instruction_rms": 0.00217352778190971, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967563085258007, + "max_absolute_post_projection_soma_slope": 1.08128768161464e-08, + "max_absolute_pre_projection_soma_slope": 0.02967563085258007, + "max_positive_post_projection_soma_slope": 2.8763282866606232e-09, + "min_post_projection_soma_slope": -1.08128768161464e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8581744343468447e-08, + "pre_projection_traffic_rms_ratio": 0.004945557596697166 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.599774360656738, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.93962860107422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0802244320511818, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7828353643417358, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16901946067810059, + "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": 0.7483190894126892, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0987572968006134, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.763838529586792, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08727656304836273, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12701134383678436, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7007901668548584, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16908571124076843, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1646714210510254, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020136289815349543, + "step": 189, + "teaching_rms": 0.0021735277500717527, + "traffic_rms": 0.02001731262879243 + }, + { + "batch_loss": 1.8048477172851562, + "innovation_rms": 0.0023009581074013522, + "instruction_rms": 0.0023009581074013522, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967560850083828, + "max_absolute_post_projection_soma_slope": 1.2833268669965037e-08, + "max_absolute_pre_projection_soma_slope": 0.02967560850083828, + "max_positive_post_projection_soma_slope": 1.2833268669965037e-08, + "min_post_projection_soma_slope": -3.736352116590069e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8603126280952476e-08, + "pre_projection_traffic_rms_ratio": 0.005315456185848757 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.616783618927002, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.25975036621094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06600309908390045, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7840596437454224, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19711266458034515, + "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": 0.7387586832046509, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10524431616067886, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7715468406677246, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09138767421245575, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14010731875896454, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6949038505554199, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19699278473854065, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1650272607803345, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02017532319450579, + "step": 190, + "teaching_rms": 0.0023009581074013522, + "traffic_rms": 0.02004596615597493 + }, + { + "batch_loss": 1.7237906455993652, + "innovation_rms": 0.0022766478484434727, + "instruction_rms": 0.0022766478462723397, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 9.796526256877769e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.496726990765865e-09, + "min_post_projection_soma_slope": -9.796526256877769e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.847467098044516e-08, + "pre_projection_traffic_rms_ratio": 0.004893147475205358 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.621551990509033, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.71404266357422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.059606682509183884, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.787236213684082, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11568483710289001, + "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": 0.7266709208488464, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12087749689817429, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7793068885803223, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08500108122825623, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10479893535375595, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6925516724586487, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11586295068264008, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165332317352295, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020169438917715186, + "step": 191, + "teaching_rms": 0.0022766478484434727, + "traffic_rms": 0.020042252492344356 + }, + { + "batch_loss": 1.7507412433624268, + "innovation_rms": 0.002203489457262296, + "instruction_rms": 0.0022034894482894263, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 3.9540073437649426e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.9540073437649426e-09, + "min_post_projection_soma_slope": -2.902890594569385e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8592571345870938e-08, + "pre_projection_traffic_rms_ratio": 0.005136238087546201 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.636018753051758, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.09687805175781, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0596487820148468, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.790706217288971, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16668763756752014, + "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": 0.7203351855278015, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12397487461566925, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7864406108856201, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10021653771400452, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07874543219804764, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6976204514503479, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1667252629995346, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1656190156936646, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02012534867143493, + "step": 192, + "teaching_rms": 0.002203489457262296, + "traffic_rms": 0.020006729075853727 + }, + { + "batch_loss": 1.803594946861267, + "innovation_rms": 0.0020408412695513997, + "instruction_rms": 0.0020408412695513997, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967563085258007, + "max_absolute_post_projection_soma_slope": 1.070171684602883e-08, + "max_absolute_pre_projection_soma_slope": 0.02967563085258007, + "max_positive_post_projection_soma_slope": 2.96585911385705e-09, + "min_post_projection_soma_slope": -1.070171684602883e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8742164659309005e-08, + "pre_projection_traffic_rms_ratio": 0.005059188361717808 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.658136367797852, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.20557403564453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058903761208057404, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7947221994400024, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21953347325325012, + "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": 0.7184186577796936, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1596645712852478, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7921104431152344, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11413171142339706, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08188413083553314, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7059208154678345, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21950888633728027, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.165932059288025, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020161615771615418, + "step": 193, + "teaching_rms": 0.0020408412695513997, + "traffic_rms": 0.02006230947269622 + }, + { + "batch_loss": 1.6399306058883667, + "innovation_rms": 0.0021482389928586592, + "instruction_rms": 0.0021482390043632126, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 5.769724253923414e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 2.5801747405296283e-09, + "min_post_projection_soma_slope": -5.769724253923414e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8679758925435202e-08, + "pre_projection_traffic_rms_ratio": 0.004993967912173679 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.67243766784668, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.33444213867188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.056277576833963394, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7988203763961792, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2085365653038025, + "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": 0.7230072617530823, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1657486855983734, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7947148084640503, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10785821825265884, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0938066840171814, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7164745330810547, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20851197838783264, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1662436723709106, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020144861507841904, + "step": 194, + "teaching_rms": 0.0021482389928586592, + "traffic_rms": 0.020028723304313375 + }, + { + "batch_loss": 1.7160954475402832, + "innovation_rms": 0.0020782802418991557, + "instruction_rms": 0.002078280239520792, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.344009152390527e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.1867810612595804e-09, + "min_post_projection_soma_slope": -4.344009152390527e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8652412255405507e-08, + "pre_projection_traffic_rms_ratio": 0.004868933695757362 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.686315536499023, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.66488647460938, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04861392080783844, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7991493940353394, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14231546223163605, + "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": 0.7272765636444092, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0934138372540474, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7977584600448608, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1560540646314621, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10502726584672928, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.729048490524292, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1422908753156662, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166439414024353, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02017678136269328, + "step": 195, + "teaching_rms": 0.0020782802418991557, + "traffic_rms": 0.02006611765446566 + }, + { + "batch_loss": 2.014594554901123, + "innovation_rms": 0.0020260136134315054, + "instruction_rms": 0.002026013608552065, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 3.937572046197602e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.937572046197602e-09, + "min_post_projection_soma_slope": -3.845158857984643e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8577820187380432e-08, + "pre_projection_traffic_rms_ratio": 0.004832110589372154 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.694983959197998, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.57502746582031, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05200977623462677, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7972685694694519, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12018800526857376, + "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": 0.7258700132369995, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07903729379177094, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7950867414474487, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1572161167860031, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06931741535663605, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7347899675369263, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12005632370710373, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1665353775024414, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020151642265777113, + "step": 196, + "teaching_rms": 0.0020260136134315054, + "traffic_rms": 0.020050790322958072 + }, + { + "batch_loss": 1.7812483310699463, + "innovation_rms": 0.0019745129086136963, + "instruction_rms": 0.0019745129011036326, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 4.308986500944911e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.308986500944911e-09, + "min_post_projection_soma_slope": -3.958510408352822e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8621486343610434e-08, + "pre_projection_traffic_rms_ratio": 0.00474996217766652 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.707380294799805, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.60626220703125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06139053404331207, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7966349124908447, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17643976211547852, + "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": 0.7228409051895142, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08627989888191223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7906876802444458, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15722227096557617, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08359051495790482, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7397626638412476, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17661190032958984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1667125225067139, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020173553513846063, + "step": 197, + "teaching_rms": 0.0019745129086136963, + "traffic_rms": 0.020075758804165816 + }, + { + "batch_loss": 1.7611521482467651, + "innovation_rms": 0.0018898750392756217, + "instruction_rms": 0.001889875040583355, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334168881177902, + "max_absolute_post_projection_soma_slope": 6.8330612279510206e-09, + "max_absolute_pre_projection_soma_slope": 0.029334168881177902, + "max_positive_post_projection_soma_slope": 6.8330612279510206e-09, + "min_post_projection_soma_slope": -3.6347662657476576e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8451290149032738e-08, + "pre_projection_traffic_rms_ratio": 0.004940233513308042 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.711256980895996, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.93358612060547, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.057580504566431046, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7993977665901184, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13243597745895386, + "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": 0.7225694060325623, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0965845137834549, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7857085466384888, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14041957259178162, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07057259976863861, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7352204322814941, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1326080858707428, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1669877767562866, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020261641661810336, + "step": 198, + "teaching_rms": 0.0018898750392756217, + "traffic_rms": 0.02017306113802941 + }, + { + "batch_loss": 1.9224169254302979, + "innovation_rms": 0.0018084608715061938, + "instruction_rms": 0.0018084608619399537, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.872063280065731e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 5.872063280065731e-09, + "min_post_projection_soma_slope": -4.827183097688703e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8432353947588357e-08, + "pre_projection_traffic_rms_ratio": 0.004582206828996866 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.718252658843994, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.17945098876953, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04937637597322464, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8005969524383545, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.150617316365242, + "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": 0.718334436416626, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08599529415369034, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7800848484039307, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14621716737747192, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07554297149181366, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7296669483184814, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15053558349609375, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1671086549758911, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020255050685265084, + "step": 199, + "teaching_rms": 0.0018084608715061938, + "traffic_rms": 0.020176764372839735 + }, + { + "batch_loss": 1.926003336906433, + "innovation_rms": 0.0015166347403684655, + "instruction_rms": 0.001516634743627593, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.4415876815074853e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.3302725022110735e-09, + "min_post_projection_soma_slope": -3.4415876815074853e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.857885065715325e-08, + "pre_projection_traffic_rms_ratio": 0.004545460188824826 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.731129169464111, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.84820556640625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06528811156749725, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8002243041992188, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19941285252571106, + "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": 0.7190173268318176, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11513651907444, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7746813297271729, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11791044473648071, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08826883882284164, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7274600267410278, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1990966647863388, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.167153239250183, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02016135603908803, + "step": 200, + "teaching_rms": 0.0015166347403684655, + "traffic_rms": 0.020107678390393065 + }, + { + "batch_loss": 1.7316089868545532, + "innovation_rms": 0.0015181848952363936, + "instruction_rms": 0.0015181848936084935, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 5.131096436628013e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 5.131096436628013e-09, + "min_post_projection_soma_slope": -3.5661549269150328e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8536953793270434e-08, + "pre_projection_traffic_rms_ratio": 0.004441755385106357 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.7459869384765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.90394592285156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09079831838607788, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.800261914730072, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1427963823080063, + "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": 0.7201005220413208, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1112336814403534, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7692011594772339, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09600017219781876, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09425682574510574, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7253080010414124, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14248022437095642, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1669741868972778, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02015155686502833, + "step": 201, + "teaching_rms": 0.0015181848952363936, + "traffic_rms": 0.020095274027789205 + }, + { + "batch_loss": 1.868704915046692, + "innovation_rms": 0.0015281709804201929, + "instruction_rms": 0.0015281709771856685, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 7.071653485013485e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 7.071653485013485e-09, + "min_post_projection_soma_slope": -4.215210847036133e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.835224048082962e-08, + "pre_projection_traffic_rms_ratio": 0.0044677687116070975 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.747967720031738, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.88565826416016, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08235874027013779, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7996602654457092, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16327890753746033, + "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": 0.7190819978713989, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11153838783502579, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7629311084747314, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10203709453344345, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08969753980636597, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7230821847915649, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16296277940273285, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1667691469192505, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020273533197279543, + "step": 202, + "teaching_rms": 0.0015281709804201929, + "traffic_rms": 0.02021774616462073 + }, + { + "batch_loss": 1.7518459558486938, + "innovation_rms": 0.0014341852707533528, + "instruction_rms": 0.0014341852673068621, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.47693851487918e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.4120843928064915e-09, + "min_post_projection_soma_slope": -3.47693851487918e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.843219078948189e-08, + "pre_projection_traffic_rms_ratio": 0.00452104274371455 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.756289958953857, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.65322875976562, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047319136559963226, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7944732904434204, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17877407371997833, + "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": 0.7203707695007324, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13034860789775848, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7565841674804688, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10970072448253632, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06299804151058197, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7116434574127197, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17845796048641205, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1666980981826782, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020214568892050495, + "step": 203, + "teaching_rms": 0.0014341852707533528, + "traffic_rms": 0.020163003575716387 + }, + { + "batch_loss": 1.7394200563430786, + "innovation_rms": 0.0015193175903045356, + "instruction_rms": 0.0015193175837977906, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 8.624232883391869e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.482362925273264e-09, + "min_post_projection_soma_slope": -8.624232883391869e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8464821922236043e-08, + "pre_projection_traffic_rms_ratio": 0.004601142996720786 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.759968280792236, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.47491455078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05611128732562065, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7899033427238464, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17134591937065125, + "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": 0.7203150391578674, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12365905195474625, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7537634372711182, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12667059898376465, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08348590135574341, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7013306021690369, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17102983593940735, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1665329933166504, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0201892357710635, + "step": 204, + "teaching_rms": 0.0015193175903045356, + "traffic_rms": 0.020130692389389825 + }, + { + "batch_loss": 1.7422103881835938, + "innovation_rms": 0.00140985441538168, + "instruction_rms": 0.0014098544171346648, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.453836576270078e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 6.453836576270078e-09, + "min_post_projection_soma_slope": -2.464501269727748e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8361151060212055e-08, + "pre_projection_traffic_rms_ratio": 0.004682280081525012 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.767888069152832, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.46261596679688, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03740008547902107, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.788695752620697, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08021364361047745, + "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": 0.7217262983322144, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09279722720384598, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7492680549621582, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14208315312862396, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12788622081279755, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6981357336044312, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10920222848653793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1662875413894653, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020201654206672193, + "step": 205, + "teaching_rms": 0.00140985441538168, + "traffic_rms": 0.02014806714918336 + }, + { + "batch_loss": 1.8238056898117065, + "innovation_rms": 0.0015705101479897325, + "instruction_rms": 0.0015705101558580452, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 8.551082508745367e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 8.551082508745367e-09, + "min_post_projection_soma_slope": -1.9328534328622027e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8525950296411084e-08, + "pre_projection_traffic_rms_ratio": 0.004393857087840166 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.769775390625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.2911605834961, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04128202050924301, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7871144413948059, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11812537908554077, + "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": 0.7220367193222046, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1117902547121048, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.741665005683899, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15678098797798157, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1392146646976471, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6909410953521729, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1181999146938324, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.166082739830017, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02014573678154518, + "step": 206, + "teaching_rms": 0.0015705101479897325, + "traffic_rms": 0.020084089667445057 + }, + { + "batch_loss": 1.7353357076644897, + "innovation_rms": 0.0016582911850061145, + "instruction_rms": 0.0016582911790446692, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.276467263115592e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.7088746801240404e-09, + "min_post_projection_soma_slope": -5.276467263115592e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.855307923377421e-08, + "pre_projection_traffic_rms_ratio": 0.004880471936037816 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.7695512771606445, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.09414672851562, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04530054330825806, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7872132062911987, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1374884843826294, + "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": 0.7243648767471313, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09383444488048553, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7353583574295044, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13756734132766724, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1397721767425537, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6818845272064209, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13756413757801056, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1656436920166016, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020153329122769768, + "step": 207, + "teaching_rms": 0.0016582911850061145, + "traffic_rms": 0.020083126219064343 + }, + { + "batch_loss": 1.641133189201355, + "innovation_rms": 0.0015044921912675583, + "instruction_rms": 0.0015044921945529899, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.5080566774279305e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 5.5080566774279305e-09, + "min_post_projection_soma_slope": -3.127975212180445e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8555458126067914e-08, + "pre_projection_traffic_rms_ratio": 0.004797582424188543 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.768247604370117, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.3479232788086, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06294699758291245, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7901157736778259, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17943629622459412, + "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": 0.7236117124557495, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11072717607021332, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.727535605430603, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14855913817882538, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1611780822277069, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6718969345092773, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17951194941997528, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1650804281234741, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020165187658135776, + "step": 208, + "teaching_rms": 0.0015044921912675583, + "traffic_rms": 0.02010706126246503 + }, + { + "batch_loss": 1.7381770610809326, + "innovation_rms": 0.0017493607411341065, + "instruction_rms": 0.0017493607439596564, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.007754128565466e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.671583925779487e-09, + "min_post_projection_soma_slope": -4.007754128565466e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8536663492753775e-08, + "pre_projection_traffic_rms_ratio": 0.004963255862966761 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.759856700897217, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.62483978271484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058326318860054016, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7932907938957214, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21367576718330383, + "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": 0.7225008606910706, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10280535370111465, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7207677364349365, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13746683299541473, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13159091770648956, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6623331904411316, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2139916867017746, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1645177602767944, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020212800930754327, + "step": 209, + "teaching_rms": 0.0017493607411341065, + "traffic_rms": 0.020136381129941434 + }, + { + "batch_loss": 1.7629002332687378, + "innovation_rms": 0.0017618840892902239, + "instruction_rms": 0.001761884097706622, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 1.4122399250027229e-08, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 1.7080876713038151e-09, + "min_post_projection_soma_slope": -1.4122399250027229e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8616331623991966e-08, + "pre_projection_traffic_rms_ratio": 0.004653560809337282 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.758052825927734, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.0620346069336, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0680479109287262, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7952477931976318, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2443057745695114, + "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": 0.7217362523078918, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09356145560741425, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.712680697441101, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13041357696056366, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14714805781841278, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6597880125045776, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2446216642856598, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1639705896377563, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020250387914674907, + "step": 210, + "teaching_rms": 0.0017618840892902239, + "traffic_rms": 0.02016917085495485 + }, + { + "batch_loss": 1.880725622177124, + "innovation_rms": 0.0017494134495598154, + "instruction_rms": 0.0017494134453216183, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.104120764405252e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.4175955399007307e-09, + "min_post_projection_soma_slope": -6.104120764405252e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.849844095735569e-08, + "pre_projection_traffic_rms_ratio": 0.0049983482786607695 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.755855083465576, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.8750228881836, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06707611680030823, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7969669103622437, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18353211879730225, + "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": 0.7170640826225281, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10416507720947266, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7030681371688843, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12379016727209091, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11651082336902618, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.662436842918396, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18384796380996704, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1633977890014648, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020275554363983452, + "step": 211, + "teaching_rms": 0.0017494134495598154, + "traffic_rms": 0.020202308174149104 + }, + { + "batch_loss": 1.6597987413406372, + "innovation_rms": 0.001732447495005118, + "instruction_rms": 0.001732447495005118, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 7.44897521443022e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.7514836509776615e-09, + "min_post_projection_soma_slope": -7.44897521443022e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.864673768963325e-08, + "pre_projection_traffic_rms_ratio": 0.004932631099708152 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.7607808113098145, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.30341339111328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06670162826776505, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7987412810325623, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14701922237873077, + "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": 0.7114233374595642, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09340979903936386, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6967743635177612, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12119165062904358, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10016734153032303, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6656672954559326, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.147335022687912, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.162712812423706, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020270917697611582, + "step": 212, + "teaching_rms": 0.001732447495005118, + "traffic_rms": 0.020197383055323485 + }, + { + "batch_loss": 1.778031587600708, + "innovation_rms": 0.0018657962381969351, + "instruction_rms": 0.0018657962434953765, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.18648574193503e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.481913341512154e-09, + "min_post_projection_soma_slope": -3.18648574193503e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8633243252113398e-08, + "pre_projection_traffic_rms_ratio": 0.004967089594296668 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.760693550109863, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.85813903808594, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04664745181798935, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7999213933944702, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11434595286846161, + "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": 0.7077215909957886, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0907215029001236, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6959238052368164, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08545452356338501, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10743864625692368, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.654819905757904, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11430840194225311, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.162116527557373, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02028455384990774, + "step": 213, + "teaching_rms": 0.0018657962381969351, + "traffic_rms": 0.020201373308182505 + }, + { + "batch_loss": 1.604391098022461, + "innovation_rms": 0.0017567562954889614, + "instruction_rms": 0.001756756294082134, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.295030553436163e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.958350092148066e-09, + "min_post_projection_soma_slope": -4.295030553436163e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8710000640865155e-08, + "pre_projection_traffic_rms_ratio": 0.005117647069915925 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.761440753936768, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.3661117553711, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05481785535812378, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8005993962287903, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22198262810707092, + "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": 0.7054536938667297, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11388710141181946, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7013518810272217, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06428654491901398, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09066595882177353, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6413916349411011, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2220860868692398, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1616359949111938, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020281043364827462, + "step": 214, + "teaching_rms": 0.0017567562954889614, + "traffic_rms": 0.020206431392375417 + }, + { + "batch_loss": 1.8421554565429688, + "innovation_rms": 0.0021542621861016095, + "instruction_rms": 0.0021542621792181773, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 7.532459989079143e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.0566940917585725e-09, + "min_post_projection_soma_slope": -7.532459989079143e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8742333154875687e-08, + "pre_projection_traffic_rms_ratio": 0.005032304734077958 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.759544372558594, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.46146392822266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06742331385612488, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7996049523353577, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25671708583831787, + "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": 0.70596843957901, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14498619735240936, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.706904411315918, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0762038379907608, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09900374710559845, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.629582941532135, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.256401389837265, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1612862348556519, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020284804595297846, + "step": 215, + "teaching_rms": 0.0021542621861016095, + "traffic_rms": 0.020175633717278416 + }, + { + "batch_loss": 1.665122628211975, + "innovation_rms": 0.0020301268900367333, + "instruction_rms": 0.0020301268827324025, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967561036348343, + "max_absolute_post_projection_soma_slope": 9.687926905144195e-09, + "max_absolute_pre_projection_soma_slope": 0.02967561036348343, + "max_positive_post_projection_soma_slope": 9.687926905144195e-09, + "min_post_projection_soma_slope": -5.910632427941209e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8764083878366167e-08, + "pre_projection_traffic_rms_ratio": 0.005344821576045932 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.759164333343506, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.57575988769531, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07568767666816711, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.79897540807724, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.254562109708786, + "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": 0.7069238424301147, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13005661964416504, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7106521129608154, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07469492405653, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09363902360200882, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6243107318878174, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25424644351005554, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.160975694656372, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020298288952060872, + "step": 216, + "teaching_rms": 0.0020301268900367333, + "traffic_rms": 0.020196303678005522 + }, + { + "batch_loss": 1.7769309282302856, + "innovation_rms": 0.0021380264698722825, + "instruction_rms": 0.002138026481431789, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.853306810938875e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 6.853306810938875e-09, + "min_post_projection_soma_slope": -4.11528544574935e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.855362332766742e-08, + "pre_projection_traffic_rms_ratio": 0.005091452480126162 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.749573230743408, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.91468048095703, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06535797566175461, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7997252941131592, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20501968264579773, + "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": 0.7110123038291931, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09627837687730789, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7115099430084229, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06920552998781204, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09456519037485123, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6219761967658997, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20470404624938965, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1607294082641602, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02027926077520831, + "step": 217, + "teaching_rms": 0.0021380264698722825, + "traffic_rms": 0.02016404237831238 + }, + { + "batch_loss": 1.8075400590896606, + "innovation_rms": 0.002147675652615635, + "instruction_rms": 0.00214767563880655, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.526815284260465e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.526815284260465e-09, + "min_post_projection_soma_slope": -2.5463526842628426e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8628998210354773e-08, + "pre_projection_traffic_rms_ratio": 0.005116066064007355 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.743312835693359, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.17644500732422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.055139172822237015, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8012726306915283, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14532631635665894, + "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": 0.719329297542572, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08317004144191742, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7109577655792236, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09813185036182404, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10952121019363403, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6289567351341248, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14553165435791016, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1605321168899536, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02027169368337891, + "step": 218, + "teaching_rms": 0.002147675652615635, + "traffic_rms": 0.020154144566732136 + }, + { + "batch_loss": 1.7020210027694702, + "innovation_rms": 0.0021860656638853773, + "instruction_rms": 0.0021860656593631835, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.438879568941957e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 5.438879568941957e-09, + "min_post_projection_soma_slope": -2.5425770378006973e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8534139442325453e-08, + "pre_projection_traffic_rms_ratio": 0.005272977771100156 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.730807304382324, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.83289337158203, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0710449144244194, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8018054366111755, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11372274160385132, + "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": 0.728691816329956, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09362504631280899, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7083706855773926, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11138682812452316, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10917018353939056, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6354793310165405, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11396923661231995, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.160508394241333, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02033332695995979, + "step": 219, + "teaching_rms": 0.0021860656638853773, + "traffic_rms": 0.020212969825867514 + }, + { + "batch_loss": 1.8164851665496826, + "innovation_rms": 0.0021674024243069855, + "instruction_rms": 0.002167402406062451, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967563085258007, + "max_absolute_post_projection_soma_slope": 9.521830435232914e-09, + "max_absolute_pre_projection_soma_slope": 0.02967563085258007, + "max_positive_post_projection_soma_slope": 6.398582552691323e-09, + "min_post_projection_soma_slope": -9.521830435232914e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8875784056382825e-08, + "pre_projection_traffic_rms_ratio": 0.0052960884293847856 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.730015277862549, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.06539916992188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09205417335033417, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8009334206581116, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15651044249534607, + "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": 0.7397376894950867, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11045857518911362, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.704600214958191, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15650923550128937, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10400306433439255, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.64694744348526, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1567569077014923, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.160566806793213, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02014789336371992, + "step": 220, + "teaching_rms": 0.0021674024243069855, + "traffic_rms": 0.020028519519173355 + }, + { + "batch_loss": 1.6777905225753784, + "innovation_rms": 0.00195946347885092, + "instruction_rms": 0.001959463471283176, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.0139608021027016e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.953472743096654e-09, + "min_post_projection_soma_slope": -5.0139608021027016e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.866657334218883e-08, + "pre_projection_traffic_rms_ratio": 0.005365669991736808 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.725189208984375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.4670181274414, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09654171019792557, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7999117970466614, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1466698795557022, + "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": 0.7462562322616577, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08941017836332321, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.701390027999878, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1738104522228241, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09652131050825119, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6554889678955078, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14680123329162598, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1605570316314697, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020214165358324054, + "step": 221, + "teaching_rms": 0.00195946347885092, + "traffic_rms": 0.020117292170537204 + }, + { + "batch_loss": 1.751305103302002, + "innovation_rms": 0.002021528349191089, + "instruction_rms": 0.0020215283467459565, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.433928779690177e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.6246847745502464e-09, + "min_post_projection_soma_slope": -4.433928779690177e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8574864019852155e-08, + "pre_projection_traffic_rms_ratio": 0.005350919954833335 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.726551532745361, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.94306182861328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08480940014123917, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8000071048736572, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19781513512134552, + "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": 0.7487424612045288, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11372260749340057, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7061688899993896, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14139193296432495, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08465104550123215, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6692213416099548, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19813062250614166, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1606881618499756, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0202599617316421, + "step": 222, + "teaching_rms": 0.002021528349191089, + "traffic_rms": 0.020158151742530002 + }, + { + "batch_loss": 1.5751615762710571, + "innovation_rms": 0.0016927258143014704, + "instruction_rms": 0.0016927258157615136, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 6.459787371682069e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 6.459787371682069e-09, + "min_post_projection_soma_slope": -3.017587957288015e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8647829878368287e-08, + "pre_projection_traffic_rms_ratio": 0.005278345192235705 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.737143516540527, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.26261901855469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0780433714389801, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7989445328712463, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15332551300525665, + "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": 0.751176118850708, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10813717544078827, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.721683144569397, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1568189412355423, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07790498435497284, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6784687042236328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1536409705877304, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.160799503326416, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020264789714215053, + "step": 223, + "teaching_rms": 0.0016927258143014704, + "traffic_rms": 0.020192325763476293 + }, + { + "batch_loss": 1.8620271682739258, + "innovation_rms": 0.002028974329061557, + "instruction_rms": 0.0020289743314977167, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967562898993492, + "max_absolute_post_projection_soma_slope": 4.920488017035041e-09, + "max_absolute_pre_projection_soma_slope": 0.02967562898993492, + "max_positive_post_projection_soma_slope": 4.5259902492489346e-09, + "min_post_projection_soma_slope": -4.920488017035041e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8467886299304433e-08, + "pre_projection_traffic_rms_ratio": 0.005314109083290234 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.742325782775879, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.48048400878906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05536094307899475, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7996971607208252, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17458969354629517, + "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": 0.7460953593254089, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14888256788253784, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.737663745880127, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15980607271194458, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06970240920782089, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6783571839332581, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17490512132644653, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.16084885597229, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020343744335589137, + "step": 224, + "teaching_rms": 0.002028974329061557, + "traffic_rms": 0.020245611446470177 + }, + { + "batch_loss": 1.7548381090164185, + "innovation_rms": 0.001977387627250406, + "instruction_rms": 0.0019773876247506906, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.274863130149242e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.274863130149242e-09, + "min_post_projection_soma_slope": -4.087214566794728e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8505412258915934e-08, + "pre_projection_traffic_rms_ratio": 0.005186071248713427 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.748894214630127, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.46038055419922, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.049905434250831604, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.801950216293335, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12039437890052795, + "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": 0.7411956787109375, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12115895003080368, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7530603408813477, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15396587550640106, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10190607607364655, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6781620979309082, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12035688012838364, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1609631776809692, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020310865269672258, + "step": 225, + "teaching_rms": 0.001977387627250406, + "traffic_rms": 0.020216795906679057 + }, + { + "batch_loss": 1.7869658470153809, + "innovation_rms": 0.0019495678109583236, + "instruction_rms": 0.0019495678185644802, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 1.0642459358223277e-08, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 5.484703802238755e-09, + "min_post_projection_soma_slope": -1.0642459358223277e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8436167200867892e-08, + "pre_projection_traffic_rms_ratio": 0.00524876092474966 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.75487756729126, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 83.03048706054688, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0695323571562767, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8007833957672119, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14662054181098938, + "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": 0.7354136109352112, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10502900183200836, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7630523443222046, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11929972469806671, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10851886868476868, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6667970418930054, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14662355184555054, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1609712839126587, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020318239707699734, + "step": 226, + "teaching_rms": 0.0019495678109583236, + "traffic_rms": 0.020227783148881715 + }, + { + "batch_loss": 1.5921454429626465, + "innovation_rms": 0.0017899698714828399, + "instruction_rms": 0.0017899698742442863, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 8.262834860772728e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 8.262834860772728e-09, + "min_post_projection_soma_slope": -2.911236585134702e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8374658989275135e-08, + "pre_projection_traffic_rms_ratio": 0.005263569314813644 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.761349201202393, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 82.32756805419922, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06222611665725708, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8007155060768127, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17470870912075043, + "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": 0.7264096140861511, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11162231117486954, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7677433490753174, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11518655717372894, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.091776043176651, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6588994264602661, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1745097041130066, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1609331369400024, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020295046309772545, + "step": 227, + "teaching_rms": 0.0017899698714828399, + "traffic_rms": 0.02021473042945786 + }, + { + "batch_loss": 1.6244009733200073, + "innovation_rms": 0.001944411515166499, + "instruction_rms": 0.001944411520250717, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.953050496998458e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 5.200187391807276e-09, + "min_post_projection_soma_slope": -5.953050496998458e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8464686184751875e-08, + "pre_projection_traffic_rms_ratio": 0.0053141994382962954 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.773193359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 81.47850036621094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06691567599773407, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8022405505180359, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16819393634796143, + "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": 0.7214599251747131, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08918315172195435, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7695223093032837, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11484260112047195, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0819382518529892, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6530506014823914, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1678786277770996, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1608914136886597, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020324079382859364, + "step": 228, + "teaching_rms": 0.001944411515166499, + "traffic_rms": 0.02023048542148319 + }, + { + "batch_loss": 1.5846376419067383, + "innovation_rms": 0.0020997942576505947, + "instruction_rms": 0.002099794252942604, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 7.039789196028323e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.631602352172081e-09, + "min_post_projection_soma_slope": -7.039789196028323e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.857483598515591e-08, + "pre_projection_traffic_rms_ratio": 0.004903527491846575 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.783061981201172, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 80.89773559570312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05920372158288956, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8033092617988586, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14120720326900482, + "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": 0.7213201522827148, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08757813274860382, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.768689751625061, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12399397045373917, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07393672317266464, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6578788757324219, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1408919245004654, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1605687141418457, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02026378927887164, + "step": 229, + "teaching_rms": 0.0020997942576505947, + "traffic_rms": 0.020155468527305487 + }, + { + "batch_loss": 1.5751243829727173, + "innovation_rms": 0.0019071043262338856, + "instruction_rms": 0.001907104328825724, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 1.0738276934318947e-08, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.075657284412614e-09, + "min_post_projection_soma_slope": -1.0738276934318947e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8300794064769136e-08, + "pre_projection_traffic_rms_ratio": 0.005165528518172943 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.7783522605896, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 79.9685287475586, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058293383568525314, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8050643801689148, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11348168551921844, + "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": 0.7223546504974365, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06865742802619934, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7681667804718018, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09369268268346786, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0776466354727745, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6677388548851013, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11345502734184265, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1601283550262451, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02026357657544936, + "step": 230, + "teaching_rms": 0.0019071043262338856, + "traffic_rms": 0.020170443890616684 + }, + { + "batch_loss": 1.702982783317566, + "innovation_rms": 0.002184270984339187, + "instruction_rms": 0.00218427096623555, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 4.233467354453069e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.233467354453069e-09, + "min_post_projection_soma_slope": -4.118771546046673e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.855972613945288e-08, + "pre_projection_traffic_rms_ratio": 0.004941283943232157 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.778977870941162, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.94294738769531, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06696275621652603, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8066310882568359, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12912285327911377, + "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": 0.7266079783439636, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07834604382514954, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7774708271026611, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13134771585464478, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07619200646877289, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6806426644325256, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1290379911661148, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.159847378730774, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02021620166746109, + "step": 231, + "teaching_rms": 0.002184270984339187, + "traffic_rms": 0.02009297198488793 + }, + { + "batch_loss": 1.5355868339538574, + "innovation_rms": 0.0017310631944639529, + "instruction_rms": 0.0017310631944639529, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.953772059224093e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.953772059224093e-09, + "min_post_projection_soma_slope": -3.1627536145606427e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.841233090387641e-08, + "pre_projection_traffic_rms_ratio": 0.005154067279577803 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.78373908996582, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 78.19752502441406, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04764508083462715, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8151878714561462, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18684524297714233, + "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": 0.7257740497589111, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08527755737304688, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7839373350143433, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09927838295698166, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07224439084529877, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6979358792304993, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18686974048614502, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1594558954238892, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020281150286677832, + "step": 232, + "teaching_rms": 0.0017310631944639529, + "traffic_rms": 0.020203393239455657 + }, + { + "batch_loss": 1.7264397144317627, + "innovation_rms": 0.001966741902348723, + "instruction_rms": 0.001966741887269248, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.906017896961657e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.828459327337441e-09, + "min_post_projection_soma_slope": -5.906017896961657e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8419118313198018e-08, + "pre_projection_traffic_rms_ratio": 0.004843913828457986 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.785225868225098, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.85515594482422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05226557329297066, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8245461583137512, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15837831795215607, + "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": 0.7277820706367493, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10709907859563828, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7887518405914307, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1035086065530777, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06184113025665283, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7126614451408386, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15840281546115875, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1590217351913452, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02034275500163816, + "step": 233, + "teaching_rms": 0.001966741902348723, + "traffic_rms": 0.020249948806667945 + }, + { + "batch_loss": 1.6530691385269165, + "innovation_rms": 0.0019461896458509704, + "instruction_rms": 0.0019461896255326793, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334159567952156, + "max_absolute_post_projection_soma_slope": 9.976516501808419e-09, + "max_absolute_pre_projection_soma_slope": 0.029334159567952156, + "max_positive_post_projection_soma_slope": 5.16721021526223e-09, + "min_post_projection_soma_slope": -9.976516501808419e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8453409730341207e-08, + "pre_projection_traffic_rms_ratio": 0.004930187889004082 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.788357257843018, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 77.27149200439453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04088226705789566, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.833634614944458, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13225652277469635, + "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": 0.7280529141426086, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07824604213237762, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7887073755264282, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09194760024547577, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.062324896454811096, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7258785963058472, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13217170536518097, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.158616542816162, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020291817341904522, + "step": 234, + "teaching_rms": 0.0019461896458509704, + "traffic_rms": 0.020203140056793585 + }, + { + "batch_loss": 1.6233199834823608, + "innovation_rms": 0.0018297539155899428, + "instruction_rms": 0.001829753912888538, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.876953229095761e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 2.993943537532573e-09, + "min_post_projection_soma_slope": -6.876953229095761e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.853180217271145e-08, + "pre_projection_traffic_rms_ratio": 0.004815721259293479 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.787994384765625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.90870666503906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.052627768367528915, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8448906540870667, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16448672115802765, + "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": 0.7263990044593811, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07251369953155518, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7888875007629395, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10254252701997757, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06624935567378998, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7423188090324402, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16440190374851227, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.158239722251892, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02025265334126144, + "step": 235, + "teaching_rms": 0.0018297539155899428, + "traffic_rms": 0.020173117134024057 + }, + { + "batch_loss": 1.81724214553833, + "innovation_rms": 0.0021494207514578673, + "instruction_rms": 0.0021494207560571587, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.7524700558531094e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 5.7524700558531094e-09, + "min_post_projection_soma_slope": -2.920442776499499e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8424544554890984e-08, + "pre_projection_traffic_rms_ratio": 0.004837788539474905 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.781116008758545, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.47111511230469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05373261868953705, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8528010249137878, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12068021297454834, + "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": 0.7270650863647461, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09990284591913223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.795027256011963, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1798582673072815, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0901031568646431, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7531455159187317, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1207556426525116, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1578826904296875, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020355629805029394, + "step": 236, + "teaching_rms": 0.0021494207514578673, + "traffic_rms": 0.02024591170212573 + }, + { + "batch_loss": 1.8150348663330078, + "innovation_rms": 0.001937268091570643, + "instruction_rms": 0.0019372680775374882, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.041649681596482e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 4.041649681596482e-09, + "min_post_projection_soma_slope": -3.934829351237568e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8548656498237306e-08, + "pre_projection_traffic_rms_ratio": 0.005197725643110155 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.777068138122559, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.9731216430664, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06552890688180923, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8546366095542908, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16884401440620422, + "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": 0.7218047976493835, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10728222876787186, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.797096610069275, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13669145107269287, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08975213766098022, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7527325749397278, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16891944408416748, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1575409173965454, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020222308019247275, + "step": 237, + "teaching_rms": 0.001937268091570643, + "traffic_rms": 0.02013396513486101 + }, + { + "batch_loss": 1.6910065412521362, + "innovation_rms": 0.002011520444855431, + "instruction_rms": 0.0020115204374835357, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967563085258007, + "max_absolute_post_projection_soma_slope": 9.698999825502597e-09, + "max_absolute_pre_projection_soma_slope": 0.02967563085258007, + "max_positive_post_projection_soma_slope": 4.307177725593192e-09, + "min_post_projection_soma_slope": -9.698999825502597e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8634438054785222e-08, + "pre_projection_traffic_rms_ratio": 0.004811448967453553 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.772308826446533, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.96013641357422, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05002078413963318, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8510292768478394, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1439562439918518, + "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": 0.7199630737304688, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07432448863983154, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7913233041763306, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12765778601169586, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10366733372211456, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7448790669441223, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14412060379981995, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1572834253311157, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02024858559320343, + "step": 238, + "teaching_rms": 0.002011520444855431, + "traffic_rms": 0.020149988037487247 + }, + { + "batch_loss": 1.6448988914489746, + "innovation_rms": 0.002021693017312024, + "instruction_rms": 0.002021693022201892, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.563917409607711e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 6.563917409607711e-09, + "min_post_projection_soma_slope": -4.253730701009317e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.852776013488688e-08, + "pre_projection_traffic_rms_ratio": 0.0048957247320312505 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.752479553222656, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.57198333740234, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.058893732726573944, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8461174368858337, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14312809705734253, + "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": 0.7217744588851929, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10101598501205444, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7830758094787598, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10450383275747299, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10968384891748428, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7305577993392944, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14321288466453552, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1571189165115356, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02027810098678221, + "step": 239, + "teaching_rms": 0.002021693017312024, + "traffic_rms": 0.020174139645910028 + }, + { + "batch_loss": 1.5097678899765015, + "innovation_rms": 0.0017358491491052104, + "instruction_rms": 0.001735849157647841, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 8.05022715155701e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 8.05022715155701e-09, + "min_post_projection_soma_slope": -3.947013826888224e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8482541371333835e-08, + "pre_projection_traffic_rms_ratio": 0.004856145846721008 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.724198341369629, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.20647430419922, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05225423723459244, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8400956392288208, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15734004974365234, + "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": 0.7273363471031189, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0853521078824997, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7764637470245361, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13385994732379913, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10746961086988449, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7148153185844421, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15742483735084534, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1569510698318481, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020245526660863822, + "step": 240, + "teaching_rms": 0.0017358491491052104, + "traffic_rms": 0.020168506460238152 + }, + { + "batch_loss": 1.7064179182052612, + "innovation_rms": 0.001956457813516635, + "instruction_rms": 0.0019564578008843514, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967562898993492, + "max_absolute_post_projection_soma_slope": 1.0551097773259244e-08, + "max_absolute_pre_projection_soma_slope": 0.02967562898993492, + "max_positive_post_projection_soma_slope": 1.0551097773259244e-08, + "min_post_projection_soma_slope": -7.000914958865678e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8569003761666313e-08, + "pre_projection_traffic_rms_ratio": 0.005086102445044249 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.70265007019043, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.96991729736328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.052194468677043915, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8400120139122009, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12857399880886078, + "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": 0.7353877425193787, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08051399886608124, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.767590880393982, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14181934297084808, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10750288516283035, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7108119130134583, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1287366896867752, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1568660736083984, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020217219005779587, + "step": 241, + "teaching_rms": 0.001956457813516635, + "traffic_rms": 0.020121003647754432 + }, + { + "batch_loss": 1.6576603651046753, + "innovation_rms": 0.0019081976260529538, + "instruction_rms": 0.0019081976260529538, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 3.6111718060283238e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.6111718060283238e-09, + "min_post_projection_soma_slope": -2.7225099952232767e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8506294318342886e-08, + "pre_projection_traffic_rms_ratio": 0.0047303891287091586 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.692652225494385, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.91596984863281, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06598654389381409, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8409527540206909, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13277855515480042, + "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": 0.7434704303741455, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08082688599824905, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.762906789779663, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12103267014026642, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0852111279964447, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7104460597038269, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13298650085926056, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1569269895553589, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02022606504187916, + "step": 242, + "teaching_rms": 0.0019081976260529538, + "traffic_rms": 0.02013376216255537 + }, + { + "batch_loss": 1.5984420776367188, + "innovation_rms": 0.0019113334179131384, + "instruction_rms": 0.0019113334127409316, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.4718749531446065e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.188467794818962e-09, + "min_post_projection_soma_slope": -5.4718749531446065e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8372615986458344e-08, + "pre_projection_traffic_rms_ratio": 0.0048728302613718156 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.673044681549072, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.18798828125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.043623074889183044, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8432888984680176, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09245665371417999, + "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": 0.7532156705856323, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09745246917009354, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7618579864501953, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09442543983459473, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06585190445184708, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.71302330493927, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09238126128911972, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1571528911590576, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020274900049046533, + "step": 243, + "teaching_rms": 0.0019113334179131384, + "traffic_rms": 0.020181876129027543 + }, + { + "batch_loss": 1.661126971244812, + "innovation_rms": 0.002082857276220185, + "instruction_rms": 0.0020828572809664593, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.815182919060135e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.815182919060135e-09, + "min_post_projection_soma_slope": -4.510634976639949e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.85280996333088e-08, + "pre_projection_traffic_rms_ratio": 0.004875032299350782 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6562347412109375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.56692504882812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.046324267983436584, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8482061624526978, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11871180683374405, + "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": 0.7665201425552368, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1330449879169464, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7630501985549927, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10518644005060196, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06418915838003159, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7194693684577942, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11902661621570587, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1572949886322021, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020259626330835338, + "step": 244, + "teaching_rms": 0.002082857276220185, + "traffic_rms": 0.020149903710058047 + }, + { + "batch_loss": 1.5747902393341064, + "innovation_rms": 0.0020851421087600395, + "instruction_rms": 0.0020851420957220873, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 3.821625238487059e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 2.303549351267975e-09, + "min_post_projection_soma_slope": -3.821625238487059e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.846658629687238e-08, + "pre_projection_traffic_rms_ratio": 0.005137098651338769 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.649474143981934, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.88859558105469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.045621953904628754, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8525472283363342, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1369294375181198, + "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": 0.7750929594039917, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08572826534509659, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7659432888031006, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1182294562458992, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05619005113840103, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7264741659164429, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13724422454833984, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.157346248626709, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020235358434679763, + "step": 245, + "teaching_rms": 0.0020851421087600395, + "traffic_rms": 0.020124702686884363 + }, + { + "batch_loss": 1.692156195640564, + "innovation_rms": 0.002204620116658435, + "instruction_rms": 0.002204620125626703, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 6.238210836784219e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 6.238210836784219e-09, + "min_post_projection_soma_slope": -5.0439790122425165e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.864719566905591e-08, + "pre_projection_traffic_rms_ratio": 0.004949552416836521 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.643632411956787, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.0558853149414, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05147712677717209, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8550774455070496, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18924760818481445, + "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": 0.792343020439148, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1725003719329834, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7698616981506348, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09977509081363678, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05744558572769165, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7289779782295227, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18917223811149597, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1575205326080322, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020270327288427144, + "step": 246, + "teaching_rms": 0.002204620116658435, + "traffic_rms": 0.02014933246316279 + }, + { + "batch_loss": 1.7213753461837769, + "innovation_rms": 0.00209755336948997, + "instruction_rms": 0.0020975533742029905, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.998196967420654e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.02044841862903e-09, + "min_post_projection_soma_slope": -4.998196967420654e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8692629252116784e-08, + "pre_projection_traffic_rms_ratio": 0.0048795439250461535 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.641814708709717, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.29293823242188, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06407327950000763, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8544802069664001, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20093542337417603, + "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": 0.809842586517334, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1749957799911499, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.770602822303772, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09612381458282471, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06392628699541092, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7298858165740967, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2011067271232605, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1574385166168213, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020270495850047608, + "step": 247, + "teaching_rms": 0.00209755336948997, + "traffic_rms": 0.020160229326393782 + }, + { + "batch_loss": 1.4882029294967651, + "innovation_rms": 0.001979838788624126, + "instruction_rms": 0.0019798387936173665, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.966231508836017e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 5.966231508836017e-09, + "min_post_projection_soma_slope": -3.358038735967739e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8759784518715444e-08, + "pre_projection_traffic_rms_ratio": 0.004665743600192618 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.643538951873779, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.51774597167969, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07414324581623077, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8561316728591919, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1593431681394577, + "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": 0.8225836753845215, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1274106800556183, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7711267471313477, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09933459013700485, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07424971461296082, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7327322959899902, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15936216711997986, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1572396755218506, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02022874424029082, + "step": 248, + "teaching_rms": 0.001979838788624126, + "traffic_rms": 0.020132572793337477 + }, + { + "batch_loss": 1.6567128896713257, + "innovation_rms": 0.002131128182964841, + "instruction_rms": 0.0021311281922423795, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 1.3192829051433819e-08, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 8.105635274091583e-09, + "min_post_projection_soma_slope": -1.3192829051433819e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8615472319428456e-08, + "pre_projection_traffic_rms_ratio": 0.005053335701691862 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.645721435546875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.66532897949219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08309338986873627, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8562204837799072, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12305640429258347, + "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": 0.8298283815383911, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0849396362900734, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.774307131767273, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11963281035423279, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08319985121488571, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7300242185592651, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12298105657100677, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1569691896438599, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020364194634629965, + "step": 249, + "teaching_rms": 0.002131128182964841, + "traffic_rms": 0.02025352981290976 + }, + { + "batch_loss": 1.7661173343658447, + "innovation_rms": 0.0020409841101453447, + "instruction_rms": 0.0020409841101453447, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 7.0417285336077384e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.252155960671189e-09, + "min_post_projection_soma_slope": -7.0417285336077384e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8636642649534773e-08, + "pre_projection_traffic_rms_ratio": 0.005057209582841263 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6515960693359375, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.33006286621094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06147628277540207, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8543123006820679, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12279990315437317, + "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": 0.8284231424331665, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06515000760555267, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7772777080535889, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08980102837085724, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06132933124899864, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.72392338514328, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12281639873981476, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1566482782363892, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020279030234013815, + "step": 250, + "teaching_rms": 0.0020409841101453447, + "traffic_rms": 0.02018182719629879 + }, + { + "batch_loss": 1.5883393287658691, + "innovation_rms": 0.001989410828329688, + "instruction_rms": 0.001989410830814296, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 7.245042343839714e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 2.6660578189563466e-09, + "min_post_projection_soma_slope": -7.245042343839714e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8592052127883243e-08, + "pre_projection_traffic_rms_ratio": 0.005002106943176662 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6580891609191895, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.09042358398438, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06390009820461273, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8488969802856445, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.219783216714859, + "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": 0.8225067853927612, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07417546212673187, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7783664464950562, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07753203809261322, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06954744458198547, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7126036882400513, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2196119725704193, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1563053131103516, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020307330387281917, + "step": 251, + "teaching_rms": 0.001989410828329688, + "traffic_rms": 0.020210114175503786 + }, + { + "batch_loss": 1.7073581218719482, + "innovation_rms": 0.002120852651166831, + "instruction_rms": 0.002120852655828075, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 2.7048958628483888e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.7048958628483888e-09, + "min_post_projection_soma_slope": -2.4230821793480573e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8715132063948696e-08, + "pre_projection_traffic_rms_ratio": 0.0051709826334505535 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.660627365112305, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.869140625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06995870172977448, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8438813090324402, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26214104890823364, + "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": 0.814151406288147, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0835537314414978, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7796103954315186, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09593992680311203, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08923037350177765, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7037698030471802, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.26196980476379395, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1559590101242065, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020346879095301362, + "step": 252, + "teaching_rms": 0.002120852651166831, + "traffic_rms": 0.020238821366912615 + }, + { + "batch_loss": 1.6446707248687744, + "innovation_rms": 0.0020691739223921094, + "instruction_rms": 0.0020691739223921094, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.2029499053010113e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.2029499053010113e-09, + "min_post_projection_soma_slope": -2.9416056257502987e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8875616431499825e-08, + "pre_projection_traffic_rms_ratio": 0.005336709112810896 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.656945705413818, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.37899017333984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05433279648423195, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8382270336151123, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23559442162513733, + "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": 0.8100977540016174, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07813262939453125, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7803586721420288, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12328008562326431, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06989862769842148, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6896644830703735, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23542319238185883, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1556954383850098, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020292656229881956, + "step": 253, + "teaching_rms": 0.0020691739223921094, + "traffic_rms": 0.02018710419777563 + }, + { + "batch_loss": 1.7498571872711182, + "innovation_rms": 0.002337369267094703, + "instruction_rms": 0.002337369267094703, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.480165933211765e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 2.946032973127899e-09, + "min_post_projection_soma_slope": -3.480165933211765e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.86638317329793e-08, + "pre_projection_traffic_rms_ratio": 0.00527748547620741 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.641303539276123, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.3789291381836, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04659520089626312, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8332750797271729, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19425350427627563, + "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": 0.802945077419281, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08740274608135223, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7778140306472778, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1049124076962471, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07257570326328278, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6794805526733398, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1941930651664734, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1554675102233887, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020340704508672244, + "step": 254, + "teaching_rms": 0.002337369267094703, + "traffic_rms": 0.020206637318677136 + }, + { + "batch_loss": 1.6792200803756714, + "innovation_rms": 0.0021992858911816467, + "instruction_rms": 0.0021992859046666767, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 1.0690034635274515e-08, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 6.5457070874686e-09, + "min_post_projection_soma_slope": -1.0690034635274515e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.865352395883178e-08, + "pre_projection_traffic_rms_ratio": 0.005307269865732828 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.622819900512695, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.37815856933594, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05314888805150986, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8319995999336243, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20148223638534546, + "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": 0.7998437881469727, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09403881430625916, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7678993940353394, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10990078747272491, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09065669775009155, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6767850518226624, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2014218121767044, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.155240774154663, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020324098218546745, + "step": 255, + "teaching_rms": 0.0021992858911816467, + "traffic_rms": 0.020198507501693996 + }, + { + "batch_loss": 1.6257938146591187, + "innovation_rms": 0.002175475408188952, + "instruction_rms": 0.0021754754172773672, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 6.361352777872753e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 6.361352777872753e-09, + "min_post_projection_soma_slope": -5.293150806551239e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8667125229673124e-08, + "pre_projection_traffic_rms_ratio": 0.005310442495629756 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.601815223693848, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.3280258178711, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08063696324825287, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8321877121925354, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1723175197839737, + "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": 0.7976353168487549, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09878373146057129, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.75611412525177, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15817582607269287, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08749083429574966, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6783519387245178, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17225711047649384, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1550685167312622, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020338087583633906, + "step": 256, + "teaching_rms": 0.002175475408188952, + "traffic_rms": 0.0202198110917671 + }, + { + "batch_loss": 1.8487211465835571, + "innovation_rms": 0.0024576476753274956, + "instruction_rms": 0.0024576476793499646, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.9228523479882824e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 3.920644697785747e-09, + "min_post_projection_soma_slope": -4.9228523479882824e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.877393433482842e-08, + "pre_projection_traffic_rms_ratio": 0.005094342496729205 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.566564559936523, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.64847564697266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06852111220359802, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8379824161529541, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1607784926891327, + "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": 0.798907995223999, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0784444510936737, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7446929216384888, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1253407895565033, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09772796928882599, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.6911094188690186, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16095136106014252, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1551003456115723, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02038478724258772, + "step": 257, + "teaching_rms": 0.0024576476753274956, + "traffic_rms": 0.020235201639126708 + }, + { + "batch_loss": 1.6348280906677246, + "innovation_rms": 0.0019657461623849106, + "instruction_rms": 0.00196574616364217, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967563085258007, + "max_absolute_post_projection_soma_slope": 7.135770196953217e-09, + "max_absolute_pre_projection_soma_slope": 0.02967563085258007, + "max_positive_post_projection_soma_slope": 4.242651119312768e-09, + "min_post_projection_soma_slope": -7.135770196953217e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8518918985913593e-08, + "pre_projection_traffic_rms_ratio": 0.0049381996315458155 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.535176753997803, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.89068603515625, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06336472928524017, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8428686857223511, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15238989889621735, + "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": 0.8020728230476379, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08415714651346207, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7366244792938232, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11036539077758789, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09536393731832504, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7019834518432617, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1525610238313675, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.15505051612854, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020366549534248168, + "step": 258, + "teaching_rms": 0.0019657461623849106, + "traffic_rms": 0.020272839937749147 + }, + { + "batch_loss": 1.7325422763824463, + "innovation_rms": 0.0018721839189006916, + "instruction_rms": 0.0018721839162605097, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 3.956476479771709e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 3.3406362121013444e-09, + "min_post_projection_soma_slope": -3.956476479771709e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.865768923731093e-08, + "pre_projection_traffic_rms_ratio": 0.005052314476983248 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.499642372131348, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.9826431274414, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06213939189910889, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8481465578079224, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25954514741897583, + "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": 0.8043674230575562, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0996338278055191, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7324668169021606, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11962666362524033, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.062277309596538544, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7111825942993164, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.259716272354126, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1551398038864136, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02031671282759875, + "step": 259, + "teaching_rms": 0.0018721839189006916, + "traffic_rms": 0.020233453836433955 + }, + { + "batch_loss": 1.5178911685943604, + "innovation_rms": 0.0018701034111193607, + "instruction_rms": 0.0018701034111193607, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.8386531165410815e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.8386531165410815e-09, + "min_post_projection_soma_slope": -4.0787835331457245e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8507396277146486e-08, + "pre_projection_traffic_rms_ratio": 0.005058601014365178 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.468080997467041, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.2305908203125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05811451002955437, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8497792482376099, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.29717081785202026, + "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": 0.8057594299316406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08302542567253113, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.732125997543335, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10760555416345596, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07675287872552872, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7125980854034424, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.297341912984848, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.155222773551941, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02036627912594943, + "step": 260, + "teaching_rms": 0.0018701034111193607, + "traffic_rms": 0.02028034501461201 + }, + { + "batch_loss": 1.5232009887695312, + "innovation_rms": 0.0018257541581623452, + "instruction_rms": 0.0018257541622233296, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 7.414276748107795e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 7.414276748107795e-09, + "min_post_projection_soma_slope": -3.5972107514936624e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8485574418381423e-08, + "pre_projection_traffic_rms_ratio": 0.005078381995367458 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.443212509155273, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.1603775024414, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05131934955716133, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8506748080253601, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2512117326259613, + "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": 0.8046088218688965, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07420685142278671, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.735396385192871, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11361066997051239, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07828045636415482, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7097560167312622, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.25138282775878906, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.15516996383667, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02036315605536671, + "step": 261, + "teaching_rms": 0.0018257541581623452, + "traffic_rms": 0.020280629942407452 + }, + { + "batch_loss": 1.5026681423187256, + "innovation_rms": 0.0018642320151009996, + "instruction_rms": 0.0018642320164267215, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 7.701428827999734e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 7.701428827999734e-09, + "min_post_projection_soma_slope": -2.798554277205767e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.852878621497241e-08, + "pre_projection_traffic_rms_ratio": 0.005075078136508427 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.423245906829834, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.92343139648438, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.060602378100156784, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8547338843345642, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2229960411787033, + "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": 0.8053946495056152, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.059860024601221085, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.735933542251587, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1191200241446495, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08398627489805222, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7128833532333374, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22316712141036987, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1549581289291382, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020358876959316444, + "step": 262, + "teaching_rms": 0.0018642320151009996, + "traffic_rms": 0.020269499596138513 + }, + { + "batch_loss": 1.6549322605133057, + "innovation_rms": 0.002066024073388934, + "instruction_rms": 0.002066024073388934, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 5.024081151105975e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 5.024081151105975e-09, + "min_post_projection_soma_slope": -1.7878086788769565e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8493094249843557e-08, + "pre_projection_traffic_rms_ratio": 0.005124869551515607 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.408934593200684, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.45360565185547, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0797712653875351, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8577377200126648, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1783333718776703, + "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": 0.8011295795440674, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06268161535263062, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7435201406478882, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10176045447587967, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08464528620243073, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7157385349273682, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17850442230701447, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1546435356140137, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02038853133035766, + "step": 263, + "teaching_rms": 0.002066024073388934, + "traffic_rms": 0.0202836183804099 + }, + { + "batch_loss": 1.6582330465316772, + "innovation_rms": 0.0020738376145788095, + "instruction_rms": 0.0020738376288795607, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.580103490387955e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.994669039997234e-09, + "min_post_projection_soma_slope": -5.580103490387955e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8407397981425668e-08, + "pre_projection_traffic_rms_ratio": 0.005360214367429172 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.3985466957092285, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.57500457763672, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0812835842370987, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8609455823898315, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11701498925685883, + "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": 0.7930348515510559, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08094757050275803, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7460911273956299, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1212783083319664, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09366056323051453, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.719364583492279, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11731855571269989, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1544140577316284, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020364651806552482, + "step": 264, + "teaching_rms": 0.0020738376145788095, + "traffic_rms": 0.020257289666215143 + }, + { + "batch_loss": 1.6427520513534546, + "innovation_rms": 0.001921748034919142, + "instruction_rms": 0.001921748034919142, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 6.791553097684755e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 6.791553097684755e-09, + "min_post_projection_soma_slope": -3.898940725832745e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8389456252417373e-08, + "pre_projection_traffic_rms_ratio": 0.005228097850115044 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.385091304779053, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.21903991699219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08685517311096191, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8594560027122498, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1183970719575882, + "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": 0.7808243036270142, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12210527062416077, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7504936456680298, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1271321177482605, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09591871500015259, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7159957885742188, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1329885572195053, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1541370153427124, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020414444109087652, + "step": 265, + "teaching_rms": 0.001921748034919142, + "traffic_rms": 0.02032391969080336 + }, + { + "batch_loss": 1.5326730012893677, + "innovation_rms": 0.001845928821353009, + "instruction_rms": 0.0018459288159975414, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 4.211992532532349e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 4.211992532532349e-09, + "min_post_projection_soma_slope": -3.025415695745437e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8419060965323795e-08, + "pre_projection_traffic_rms_ratio": 0.004971946269019024 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.3754425048828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.55290222167969, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07892314344644547, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8577846884727478, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12976887822151184, + "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": 0.7679735422134399, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14985238015651703, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7552789449691772, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10374028980731964, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10009495913982391, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7135747671127319, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12973150610923767, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1540069580078125, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0203966526914949, + "step": 266, + "teaching_rms": 0.001845928821353009, + "traffic_rms": 0.020312674024859253 + }, + { + "batch_loss": 1.6874128580093384, + "innovation_rms": 0.00213554656025845, + "instruction_rms": 0.002135546546370935, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 7.030221294002104e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 7.030221294002104e-09, + "min_post_projection_soma_slope": -2.215535976901606e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8503194716692738e-08, + "pre_projection_traffic_rms_ratio": 0.005191040484451396 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.372784614562988, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.04338073730469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05327550321817398, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8560618758201599, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19157996773719788, + "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": 0.7575273513793945, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10446184873580933, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.756659746170044, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1287478655576706, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10550853610038757, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.709553062915802, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1915196180343628, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1539641618728638, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020459548164502083, + "step": 267, + "teaching_rms": 0.00213554656025845, + "traffic_rms": 0.02034914438495779 + }, + { + "batch_loss": 1.7412450313568115, + "innovation_rms": 0.0019722458847595634, + "instruction_rms": 0.0019722458672159385, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 7.70896591006931e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 2.8143334329655545e-09, + "min_post_projection_soma_slope": -7.70896591006931e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8418411359549788e-08, + "pre_projection_traffic_rms_ratio": 0.00526644255525876 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.369286060333252, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.64900970458984, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05587942898273468, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8583608865737915, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18200832605361938, + "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": 0.7471045255661011, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1042281985282898, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7581965923309326, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12455311417579651, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12098605185747147, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7117741107940674, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18188941478729248, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1539500951766968, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020437854076257055, + "step": 268, + "teaching_rms": 0.0019722458847595634, + "traffic_rms": 0.020343636542445004 + }, + { + "batch_loss": 1.6173431873321533, + "innovation_rms": 0.0018407511357171032, + "instruction_rms": 0.0018407511343744702, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 3.877625331938361e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 3.877625331938361e-09, + "min_post_projection_soma_slope": -3.6336949005288943e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8409438788507583e-08, + "pre_projection_traffic_rms_ratio": 0.0053501134010136816 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.382547855377197, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.2082748413086, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05372018739581108, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.858564019203186, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2008248120546341, + "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": 0.7314843535423279, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15620169043540955, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7610090970993042, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14191126823425293, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13661997020244598, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7114483118057251, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20070591568946838, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1538193225860596, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020498114180471788, + "step": 269, + "teaching_rms": 0.0018407511357171032, + "traffic_rms": 0.02041666282222298 + }, + { + "batch_loss": 1.589686393737793, + "innovation_rms": 0.0018981415305726586, + "instruction_rms": 0.0018981415266665436, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.108116957723951e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 2.2540584954100495e-09, + "min_post_projection_soma_slope": -4.108116957723951e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8361630605924864e-08, + "pre_projection_traffic_rms_ratio": 0.005288731959607345 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.3978142738342285, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.18885803222656, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.060711268335580826, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8594644069671631, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.126267671585083, + "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": 0.7220364809036255, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09447894990444183, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7590924501419067, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13977082073688507, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11345864832401276, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7107958197593689, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12609675526618958, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1537567377090454, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02042022566532204, + "step": 270, + "teaching_rms": 0.0018981415305726586, + "traffic_rms": 0.02033268694124677 + }, + { + "batch_loss": 1.7573845386505127, + "innovation_rms": 0.0019164740120838906, + "instruction_rms": 0.0019164740120838906, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 7.525748912939889e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.526293118090052e-09, + "min_post_projection_soma_slope": -7.525748912939889e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8331328062228496e-08, + "pre_projection_traffic_rms_ratio": 0.0051887781295308875 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.4249467849731445, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.28412628173828, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05618022754788399, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.860927939414978, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1045459732413292, + "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": 0.7187409996986389, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10210541635751724, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7596948146820068, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1112755686044693, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1359184980392456, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7156842947006226, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10457871109247208, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1535784006118774, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020429295516485438, + "step": 271, + "teaching_rms": 0.0019164740120838906, + "traffic_rms": 0.020340246486478655 + }, + { + "batch_loss": 1.572795033454895, + "innovation_rms": 0.0018696187773750322, + "instruction_rms": 0.0018696187786969344, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967563085258007, + "max_absolute_post_projection_soma_slope": 9.010681978338653e-09, + "max_absolute_pre_projection_soma_slope": 0.02967563085258007, + "max_positive_post_projection_soma_slope": 4.183771107335588e-09, + "min_post_projection_soma_slope": -9.010681978338653e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.843425337590302e-08, + "pre_projection_traffic_rms_ratio": 0.004962107440109632 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.451828956604004, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.33622741699219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05225775018334389, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8653131723403931, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08476884663105011, + "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": 0.7255792021751404, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06838174909353256, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7584471702575684, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0996328741312027, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08644884079694748, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7219589948654175, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.084801584482193, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1534525156021118, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020407028960819876, + "step": 272, + "teaching_rms": 0.0018696187773750322, + "traffic_rms": 0.020318613143488973 + }, + { + "batch_loss": 1.669553518295288, + "innovation_rms": 0.002086196541361464, + "instruction_rms": 0.002086196527145432, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.0293341726064682, + "max_absolute_post_projection_soma_slope": 8.90898199656931e-09, + "max_absolute_pre_projection_soma_slope": 0.0293341726064682, + "max_positive_post_projection_soma_slope": 4.681087961699859e-09, + "min_post_projection_soma_slope": -8.90898199656931e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.847607887404253e-08, + "pre_projection_traffic_rms_ratio": 0.005316941904060748 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.491667747497559, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.79222106933594, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0535711869597435, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8699179291725159, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14792147278785706, + "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": 0.7269600033760071, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08126088976860046, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.754625916481018, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13409574329853058, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06160678714513779, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7304380536079407, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14795881509780884, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1534883975982666, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02044739554403293, + "step": 273, + "teaching_rms": 0.002086196541361464, + "traffic_rms": 0.020339687534278145 + }, + { + "batch_loss": 1.8077744245529175, + "innovation_rms": 0.0021099131879990765, + "instruction_rms": 0.0021099131879990765, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 6.570143096240599e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 6.570143096240599e-09, + "min_post_projection_soma_slope": -3.998491759915623e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8389142631705257e-08, + "pre_projection_traffic_rms_ratio": 0.0055009713145897365 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.531785011291504, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.7679443359375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05264561250805855, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8722706437110901, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1279836744070053, + "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": 0.730465829372406, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09019315242767334, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7543683052062988, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08168058097362518, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05678870156407356, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7329888939857483, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1280210018157959, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1534978151321411, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02049943464414826, + "step": 274, + "teaching_rms": 0.0021099131879990765, + "traffic_rms": 0.02038772321051862 + }, + { + "batch_loss": 1.4904601573944092, + "innovation_rms": 0.0019498687646114501, + "instruction_rms": 0.0019498687646114501, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 6.910624072986593e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.6443070783320763e-09, + "min_post_projection_soma_slope": -6.910624072986593e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.834496580170184e-08, + "pre_projection_traffic_rms_ratio": 0.005264129042037313 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.576565742492676, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.21339416503906, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04417979717254639, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8732017278671265, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13861824572086334, + "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": 0.734427273273468, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08944831043481827, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7539868354797363, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07490131258964539, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04769575968384743, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7349284887313843, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1387489140033722, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1535173654556274, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02047280724663597, + "step": 275, + "teaching_rms": 0.0019498687646114501, + "traffic_rms": 0.020377864761700358 + }, + { + "batch_loss": 1.6288955211639404, + "innovation_rms": 0.0018765182043062915, + "instruction_rms": 0.0018765181937699567, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.558734723048019e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.558734723048019e-09, + "min_post_projection_soma_slope": -3.6155702876072837e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8215504621869544e-08, + "pre_projection_traffic_rms_ratio": 0.0050669089374569065 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.6201701164245605, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.25839233398438, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04739315062761307, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8718510866165161, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1508806347846985, + "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": 0.7393901944160461, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11524835228919983, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7521158456802368, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13681331276893616, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07469141483306885, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7333282232284546, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15091796219348907, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1534533500671387, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02049977592534481, + "step": 276, + "teaching_rms": 0.0018765182043062915, + "traffic_rms": 0.020411520803737565 + }, + { + "batch_loss": 1.5456339120864868, + "innovation_rms": 0.0019955807948600895, + "instruction_rms": 0.0019955807824754593, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.356430327590033e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 3.4691003403253262e-09, + "min_post_projection_soma_slope": -4.356430327590033e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.830368556317185e-08, + "pre_projection_traffic_rms_ratio": 0.005100960022880978 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.657954216003418, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.1650161743164, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05276915803551674, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8701964616775513, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1525401473045349, + "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": 0.745589554309845, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08483780920505524, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7552003860473633, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19610503315925598, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08121109008789062, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.730434238910675, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1525774598121643, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1534935235977173, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020604376588376695, + "step": 277, + "teaching_rms": 0.0019955807948600895, + "traffic_rms": 0.020503708670557255 + }, + { + "batch_loss": 1.6621347665786743, + "innovation_rms": 0.001932475204807013, + "instruction_rms": 0.001932475204807013, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.9932845353216635e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.6321812224571204e-09, + "min_post_projection_soma_slope": -5.9932845353216635e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8250538640845584e-08, + "pre_projection_traffic_rms_ratio": 0.0055837066821479734 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.689781665802002, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.80082702636719, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04265298694372177, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.870455801486969, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13348612189292908, + "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": 0.7526043653488159, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07014820724725723, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7635204792022705, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15328118205070496, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08654527366161346, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7329719662666321, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13352343440055847, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1536030769348145, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02062997918679041, + "step": 278, + "teaching_rms": 0.001932475204807013, + "traffic_rms": 0.020538831463011532 + }, + { + "batch_loss": 1.6569561958312988, + "innovation_rms": 0.0022303551441489804, + "instruction_rms": 0.0022303551441489804, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.17203835315172e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 5.17203835315172e-09, + "min_post_projection_soma_slope": -3.564347039741733e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8294235794671747e-08, + "pre_projection_traffic_rms_ratio": 0.005321435352670581 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.734576225280762, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.68002319335938, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04576832056045532, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8750692009925842, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1881723701953888, + "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": 0.7656763792037964, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1307201087474823, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7718018293380737, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11809927970170975, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07101985067129135, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.745407223701477, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18806958198547363, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1537816524505615, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020729740866191327, + "step": 279, + "teaching_rms": 0.0022303551441489804, + "traffic_rms": 0.020606494705830657 + }, + { + "batch_loss": 1.8874046802520752, + "innovation_rms": 0.002260944828641169, + "instruction_rms": 0.002260944824268744, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 7.701800086579169e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 7.701800086579169e-09, + "min_post_projection_soma_slope": -3.807990811566242e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8332221475047244e-08, + "pre_projection_traffic_rms_ratio": 0.005469472185560394 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.7782979011535645, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.67494201660156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05878232419490814, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8774589896202087, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1895231008529663, + "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": 0.7820428609848022, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16366487741470337, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7808339595794678, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1209857165813446, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.062339652329683304, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7520631551742554, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18942031264305115, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1539949178695679, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02072915202995122, + "step": 280, + "teaching_rms": 0.002260944828641169, + "traffic_rms": 0.020605473255888962 + }, + { + "batch_loss": 1.6005939245224, + "innovation_rms": 0.002067831116766821, + "instruction_rms": 0.002067831121547585, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.020893561498042e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.086554288880052e-09, + "min_post_projection_soma_slope": -6.020893561498042e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8250116137992413e-08, + "pre_projection_traffic_rms_ratio": 0.0051457852674674474 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.822843551635742, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.66529846191406, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04791118949651718, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8785282373428345, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14418074488639832, + "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": 0.7963058948516846, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1426304280757904, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7883917093276978, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09325866401195526, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04793907701969147, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7575917840003967, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14407795667648315, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1541318893432617, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02077492190780449, + "step": 281, + "teaching_rms": 0.002067831116766821, + "traffic_rms": 0.020674482442645293 + }, + { + "batch_loss": 1.670373797416687, + "innovation_rms": 0.0021254579269161886, + "instruction_rms": 0.0021254579269161886, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334168881177902, + "max_absolute_post_projection_soma_slope": 9.93791982040193e-09, + "max_absolute_pre_projection_soma_slope": 0.029334168881177902, + "max_positive_post_projection_soma_slope": 9.93791982040193e-09, + "min_post_projection_soma_slope": -5.855937956766866e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8357561622974396e-08, + "pre_projection_traffic_rms_ratio": 0.005244341852177477 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.87404203414917, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.10765075683594, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047566208988428116, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.881792426109314, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1119307428598404, + "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": 0.8025010824203491, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08296370506286621, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7888175249099731, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09245596826076508, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06016138195991516, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7643771171569824, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11213785409927368, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1540995836257935, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020786835539125994, + "step": 282, + "teaching_rms": 0.0021254579269161886, + "traffic_rms": 0.02068148671059334 + }, + { + "batch_loss": 1.6094049215316772, + "innovation_rms": 0.002195497146091649, + "instruction_rms": 0.002195497146091649, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.1592180244929295e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 6.1592180244929295e-09, + "min_post_projection_soma_slope": -2.9846065618954754e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.843270120513657e-08, + "pre_projection_traffic_rms_ratio": 0.005636672602787727 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.923696517944336, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.35419464111328, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.041620273143053055, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8874654173851013, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1301403045654297, + "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": 0.8032720685005188, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04842490330338478, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7847778797149658, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10453459620475769, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06049022823572159, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7739843726158142, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12996962666511536, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1542102098464966, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020789315641033822, + "step": 283, + "teaching_rms": 0.002195497146091649, + "traffic_rms": 0.020675598984933187 + }, + { + "batch_loss": 1.6522657871246338, + "innovation_rms": 0.002153197726124198, + "instruction_rms": 0.002153197726124198, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.64290828006142e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.8279838470979257e-09, + "min_post_projection_soma_slope": -4.64290828006142e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.840086674185119e-08, + "pre_projection_traffic_rms_ratio": 0.005497388946648412 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 7.973017692565918, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.0547866821289, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04967616870999336, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8877547383308411, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18517355620861053, + "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": 0.797554612159729, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05717458203434944, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7782506942749023, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09469117224216461, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07232747972011566, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.775441586971283, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18523380160331726, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1543450355529785, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020830057914358983, + "step": 284, + "teaching_rms": 0.002153197726124198, + "traffic_rms": 0.020721514597207828 + }, + { + "batch_loss": 1.7365251779556274, + "innovation_rms": 0.002276536807464447, + "instruction_rms": 0.0022765368031219682, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675612226128578, + "max_absolute_post_projection_soma_slope": 8.509978499660065e-09, + "max_absolute_pre_projection_soma_slope": 0.029675612226128578, + "max_positive_post_projection_soma_slope": 8.509978499660065e-09, + "min_post_projection_soma_slope": -4.380556362093557e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.831591939879925e-08, + "pre_projection_traffic_rms_ratio": 0.005529547762577792 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.0206298828125, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.49182891845703, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06414256989955902, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8837609887123108, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15937785804271698, + "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": 0.7913646697998047, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07926996052265167, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7705276012420654, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12165474146604538, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09852252900600433, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7695640325546265, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1594381034374237, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1545939445495605, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020801629661936074, + "step": 285, + "teaching_rms": 0.002276536807464447, + "traffic_rms": 0.02067816894041466 + }, + { + "batch_loss": 1.584603190422058, + "innovation_rms": 0.0018871521065549682, + "instruction_rms": 0.0018871521065549682, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334161430597305, + "max_absolute_post_projection_soma_slope": 3.9056615719346155e-09, + "max_absolute_pre_projection_soma_slope": 0.029334161430597305, + "max_positive_post_projection_soma_slope": 2.7997431040205356e-09, + "min_post_projection_soma_slope": -3.9056615719346155e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.835886940212598e-08, + "pre_projection_traffic_rms_ratio": 0.005531295596436067 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.070958137512207, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.90670013427734, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06790780276060104, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8757686614990234, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12495899200439453, + "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": 0.7832108736038208, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09360820055007935, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.761284589767456, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11003082990646362, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10804460942745209, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7570596933364868, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1250433772802353, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1548341512680054, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020765338467550676, + "step": 286, + "teaching_rms": 0.0018871521065549682, + "traffic_rms": 0.0206802261353362 + }, + { + "batch_loss": 1.633284330368042, + "innovation_rms": 0.0019960327517171052, + "instruction_rms": 0.001996032761622566, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967562898993492, + "max_absolute_post_projection_soma_slope": 7.028022164234926e-09, + "max_absolute_pre_projection_soma_slope": 0.02967562898993492, + "max_positive_post_projection_soma_slope": 4.135964015716809e-09, + "min_post_projection_soma_slope": -7.028022164234926e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8281242525047512e-08, + "pre_projection_traffic_rms_ratio": 0.005495503289179769 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.122373580932617, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.96253967285156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05715307593345642, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8680461049079895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1189991906285286, + "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": 0.7753883004188538, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11190008372068405, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7561771869659424, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11033163964748383, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08506160974502563, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.745703935623169, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11905942857265472, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1551424264907837, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020803723010795635, + "step": 287, + "teaching_rms": 0.0019960327517171052, + "traffic_rms": 0.02070741683187542 + }, + { + "batch_loss": 1.5632386207580566, + "innovation_rms": 0.0020201560022931994, + "instruction_rms": 0.0020201560022931994, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 5.8784395129407585e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 5.8784395129407585e-09, + "min_post_projection_soma_slope": -3.0667501871306513e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.830828930299916e-08, + "pre_projection_traffic_rms_ratio": 0.005694106554471829 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.173379898071289, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.5960922241211, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.042380303144454956, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8649238348007202, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11038044840097427, + "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": 0.7668096423149109, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08578665554523468, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7526906728744507, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1289394348859787, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07930221408605576, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7414550185203552, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1104406788945198, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1553101539611816, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02080673076135362, + "step": 288, + "teaching_rms": 0.0020201560022931994, + "traffic_rms": 0.020709939869288655 + }, + { + "batch_loss": 1.7191969156265259, + "innovation_rms": 0.0022571703695771174, + "instruction_rms": 0.0022571703892859322, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 4.603502468114584e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.470741554740698e-09, + "min_post_projection_soma_slope": -4.603502468114584e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8316680162798123e-08, + "pre_projection_traffic_rms_ratio": 0.005512055458362845 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.235797882080078, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.61564636230469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.041994236409664154, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8636206388473511, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1408328115940094, + "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": 0.7631188035011292, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08646352589130402, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7512478828430176, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18578995764255524, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08851689100265503, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7386114597320557, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14066222310066223, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1551785469055176, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020761084421676377, + "step": 289, + "teaching_rms": 0.0022571703695771174, + "traffic_rms": 0.02063857660941698 + }, + { + "batch_loss": 1.5047216415405273, + "innovation_rms": 0.0020531040213373104, + "instruction_rms": 0.002053104016522254, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.879758457894013e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 5.879758457894013e-09, + "min_post_projection_soma_slope": -4.953271126595382e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.825482304049388e-08, + "pre_projection_traffic_rms_ratio": 0.005499454785200455 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.288955688476562, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.47867584228516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05225151404738426, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8667699694633484, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1277591735124588, + "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": 0.7627852559089661, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06913375854492188, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.753693699836731, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16519732773303986, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08925670385360718, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7415282726287842, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12785203754901886, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.154792308807373, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020805328998371477, + "step": 290, + "teaching_rms": 0.0020531040213373104, + "traffic_rms": 0.020701075593412867 + }, + { + "batch_loss": 1.6286430358886719, + "innovation_rms": 0.002047136518934379, + "instruction_rms": 0.0020471365141052863, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.4778754038077295e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 4.4778754038077295e-09, + "min_post_projection_soma_slope": -3.9022527431598064e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.824434654933864e-08, + "pre_projection_traffic_rms_ratio": 0.005568898801821332 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.339204788208008, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.1982650756836, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05534951016306877, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8701369166374207, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12622271478176117, + "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": 0.7630513310432434, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09527620673179626, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7549223899841309, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1357211023569107, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09046979993581772, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7454367876052856, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1262182593345642, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1541826725006104, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020763047749493984, + "step": 291, + "teaching_rms": 0.002047136518934379, + "traffic_rms": 0.02066236446477199 + }, + { + "batch_loss": 1.4927146434783936, + "innovation_rms": 0.002098460399186473, + "instruction_rms": 0.0020984604156749145, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334163293242455, + "max_absolute_post_projection_soma_slope": 4.869756153880189e-09, + "max_absolute_pre_projection_soma_slope": 0.029334163293242455, + "max_positive_post_projection_soma_slope": 4.869756153880189e-09, + "min_post_projection_soma_slope": -4.58864546359905e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8486734166911406e-08, + "pre_projection_traffic_rms_ratio": 0.005274783758050363 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.385360717773438, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.32315826416016, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05162369832396507, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8740007877349854, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16042494773864746, + "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": 0.7621332406997681, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09402016550302505, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7540502548217773, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10787837952375412, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08381102234125137, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7508503198623657, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16054357588291168, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1535216569900513, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020696270166544687, + "step": 292, + "teaching_rms": 0.002098460399186473, + "traffic_rms": 0.020587309357456854 + }, + { + "batch_loss": 1.5915312767028809, + "innovation_rms": 0.0020647407113607713, + "instruction_rms": 0.0020647407161486905, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 3.5750671312229088e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 2.78034462120047e-09, + "min_post_projection_soma_slope": -3.5750671312229088e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.822443415629482e-08, + "pre_projection_traffic_rms_ratio": 0.005281006315289757 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.41960334777832, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.24665069580078, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.038977839052677155, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8775950074195862, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1531941443681717, + "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": 0.7604973912239075, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10869850218296051, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7556565999984741, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09755266457796097, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07517854869365692, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.755172848701477, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15318967401981354, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1531497240066528, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020734927435933474, + "step": 293, + "teaching_rms": 0.0020647407113607713, + "traffic_rms": 0.02063209270751165 + }, + { + "batch_loss": 1.5312365293502808, + "innovation_rms": 0.0020618796374237674, + "instruction_rms": 0.00206187964221833, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.436855466887323e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.1900058150569066e-09, + "min_post_projection_soma_slope": -3.436855466887323e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8516909644244843e-08, + "pre_projection_traffic_rms_ratio": 0.005294221965306163 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.454408645629883, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.59239196777344, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04028540477156639, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8790026307106018, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14336471259593964, + "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": 0.7591902017593384, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.101302370429039, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.753959059715271, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11287042498588562, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06934048980474472, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7566637396812439, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14338907599449158, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1526997089385986, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020729027546232372, + "step": 294, + "teaching_rms": 0.0020618796374237674, + "traffic_rms": 0.02062638871758655 + }, + { + "batch_loss": 1.570555329322815, + "innovation_rms": 0.0020400788298459973, + "instruction_rms": 0.0020400788250001987, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.038442164537969e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.038442164537969e-09, + "min_post_projection_soma_slope": -2.2136852351195557e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.825878696102729e-08, + "pre_projection_traffic_rms_ratio": 0.005328888330417203 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.475261688232422, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.90254974365234, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04249398410320282, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8775743842124939, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13155445456504822, + "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": 0.7577836513519287, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08519499003887177, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.751853585243225, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0934491753578186, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06493391841650009, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7545250654220581, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13149426877498627, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1522917747497559, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02077397391056905, + "step": 295, + "teaching_rms": 0.0020400788298459973, + "traffic_rms": 0.02067373051369881 + }, + { + "batch_loss": 1.6348352432250977, + "innovation_rms": 0.002069474281707466, + "instruction_rms": 0.002069474281707466, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.357074701033525e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 4.357074701033525e-09, + "min_post_projection_soma_slope": -2.9132227741257566e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8292243433489993e-08, + "pre_projection_traffic_rms_ratio": 0.005119639358707316 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.485930442810059, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.78005981445312, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04907569661736488, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8747723698616028, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15092113614082336, + "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": 0.755555272102356, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09317968785762787, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.749362587928772, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12224727123975754, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07835099846124649, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7491233348846436, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1510915756225586, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1518369913101196, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020761054087700875, + "step": 296, + "teaching_rms": 0.002069474281707466, + "traffic_rms": 0.02065963983173235 + }, + { + "batch_loss": 1.750643253326416, + "innovation_rms": 0.0021219983155901204, + "instruction_rms": 0.0021219983249075756, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 6.370628025109681e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 6.370628025109681e-09, + "min_post_projection_soma_slope": -3.886561739108174e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.833371827139471e-08, + "pre_projection_traffic_rms_ratio": 0.005284446588587815 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.497198104858398, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.01885223388672, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04883404076099396, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8740949034690857, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20548726618289948, + "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": 0.7546150088310242, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1111598014831543, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.746596336364746, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08022959530353546, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06741990149021149, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.745863139629364, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2056964635848999, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1514508724212646, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020716375712471945, + "step": 297, + "teaching_rms": 0.0021219983155901204, + "traffic_rms": 0.020609236433356378 + }, + { + "batch_loss": 1.6242191791534424, + "innovation_rms": 0.002040150757195639, + "instruction_rms": 0.002040150754772825, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.775441458411024e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 2.358952810865844e-09, + "min_post_projection_soma_slope": -5.775441458411024e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.829273449630452e-08, + "pre_projection_traffic_rms_ratio": 0.005441751434446439 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.49746322631836, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.76712799072266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05596040189266205, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.876319408416748, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23037433624267578, + "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": 0.7512591481208801, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09282900393009186, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.742295265197754, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09005269408226013, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07838670909404755, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7486605644226074, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23058351874351501, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1511409282684326, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020728821002637524, + "step": 298, + "teaching_rms": 0.002040150757195639, + "traffic_rms": 0.02063080364786652 + }, + { + "batch_loss": 1.7735739946365356, + "innovation_rms": 0.00214149721729439, + "instruction_rms": 0.0021414972380677786, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 9.197872685717812e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 9.197872685717812e-09, + "min_post_projection_soma_slope": -6.1494738190503995e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8544450146113314e-08, + "pre_projection_traffic_rms_ratio": 0.005336965696031196 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.508585929870605, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.83282470703125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.051285646855831146, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8822171092033386, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.190315842628479, + "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": 0.749862551689148, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11793791502714157, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7337645292282104, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08530780673027039, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06821271032094955, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7575584053993225, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19019728899002075, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1506258249282837, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020611845601423418, + "step": 299, + "teaching_rms": 0.00214149721729439, + "traffic_rms": 0.020501727640974448 + }, + { + "batch_loss": 1.6004129648208618, + "innovation_rms": 0.0020180566312732267, + "instruction_rms": 0.002018056628823887, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 3.784210278467981e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 2.8419042674698858e-09, + "min_post_projection_soma_slope": -3.784210278467981e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8299018583719292e-08, + "pre_projection_traffic_rms_ratio": 0.005189268753539123 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.505046844482422, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.02455139160156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04254188761115074, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8894578218460083, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15014132857322693, + "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": 0.7531619071960449, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10096798837184906, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7238227128982544, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09941857308149338, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05664684623479843, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7725641131401062, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15005706250667572, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1500661373138428, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020611564138561432, + "step": 300, + "teaching_rms": 0.0020180566312732267, + "traffic_rms": 0.02051060319175531 + }, + { + "batch_loss": 1.6791800260543823, + "innovation_rms": 0.001968469520350953, + "instruction_rms": 0.001968469522861993, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 4.726966817969469e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 4.726966817969469e-09, + "min_post_projection_soma_slope": -3.590962638355677e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8441699286723608e-08, + "pre_projection_traffic_rms_ratio": 0.005423967329661213 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.497466087341309, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.97618865966797, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.046215128153562546, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8974193930625916, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17402613162994385, + "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": 0.755481481552124, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08240386843681335, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7153851985931396, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08773045986890793, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0663018524646759, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7899582982063293, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17394188046455383, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1494840383529663, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020602007149778088, + "step": 301, + "teaching_rms": 0.001968469520350953, + "traffic_rms": 0.020508386058941844 + }, + { + "batch_loss": 1.6445142030715942, + "innovation_rms": 0.0019009984691881457, + "instruction_rms": 0.0019009984561873302, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967562898993492, + "max_absolute_post_projection_soma_slope": 5.193012242443729e-09, + "max_absolute_pre_projection_soma_slope": 0.02967562898993492, + "max_positive_post_projection_soma_slope": 3.0141087403734446e-09, + "min_post_projection_soma_slope": -5.193012242443729e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8461259526237875e-08, + "pre_projection_traffic_rms_ratio": 0.005309767794980311 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.483902931213379, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.8816909790039, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05720604583621025, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9050009846687317, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24100527167320251, + "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": 0.7559227347373962, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0877288430929184, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7042125463485718, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11239561438560486, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08475679159164429, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8082259893417358, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24090272188186646, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.148905873298645, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020617458691857275, + "step": 302, + "teaching_rms": 0.0019009984691881457, + "traffic_rms": 0.02052885259737296 + }, + { + "batch_loss": 1.6418763399124146, + "innovation_rms": 0.0020418692937549943, + "instruction_rms": 0.0020418693034380936, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.032808836347158e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 5.032808836347158e-09, + "min_post_projection_soma_slope": -2.051621761367528e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8515976248906168e-08, + "pre_projection_traffic_rms_ratio": 0.0053055400983303145 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.461668014526367, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.25981903076172, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.049373481422662735, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9090941548347473, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24297340214252472, + "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": 0.7580758333206177, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10036763548851013, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6949756145477295, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10067369043827057, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09448018670082092, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8174241185188293, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.24287085235118866, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1484659910202026, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020592623246958475, + "step": 303, + "teaching_rms": 0.0020418692937549943, + "traffic_rms": 0.020489820787728177 + }, + { + "batch_loss": 1.6256415843963623, + "innovation_rms": 0.0019481778785874458, + "instruction_rms": 0.0019481778684386681, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.588123214688267e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 4.588123214688267e-09, + "min_post_projection_soma_slope": -4.139624198984393e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8377004690297836e-08, + "pre_projection_traffic_rms_ratio": 0.00542575406172535 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.431934356689453, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.13798522949219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03725847601890564, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9054040908813477, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14127632975578308, + "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": 0.7620289325714111, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09105108678340912, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6898075342178345, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11609753966331482, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07192254066467285, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8144189119338989, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14143961668014526, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.148089051246643, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020584381036529528, + "step": 304, + "teaching_rms": 0.0019481778785874458, + "traffic_rms": 0.020489121486226575 + }, + { + "batch_loss": 1.704606533050537, + "innovation_rms": 0.00197196887653215, + "instruction_rms": 0.0019719688702656895, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.908375649317122e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 6.908375649317122e-09, + "min_post_projection_soma_slope": -3.546534843579252e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8267447246389693e-08, + "pre_projection_traffic_rms_ratio": 0.005642963273124731 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.399068832397461, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.08212280273438, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04011901468038559, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8944810628890991, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1686503291130066, + "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": 0.767864465713501, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0933946892619133, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6869785785675049, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08330991119146347, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05923628807067871, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7996723055839539, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16865333914756775, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.147828459739685, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02058900733872239, + "step": 305, + "teaching_rms": 0.00197196887653215, + "traffic_rms": 0.020494541936595913 + }, + { + "batch_loss": 1.7061917781829834, + "innovation_rms": 0.001970533777305312, + "instruction_rms": 0.0019705337685258784, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 9.16822973096032e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 9.16822973096032e-09, + "min_post_projection_soma_slope": -2.8891697922972526e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8341116422292176e-08, + "pre_projection_traffic_rms_ratio": 0.00530100074691724 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.37350082397461, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.05997467041016, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03454267606139183, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8813754320144653, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19966544210910797, + "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": 0.7733679413795471, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08180885761976242, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6879446506500244, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09731759130954742, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06628219783306122, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7799165844917297, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19966843724250793, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.147485613822937, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020586988415429934, + "step": 306, + "teaching_rms": 0.001970533777305312, + "traffic_rms": 0.020492971448618082 + }, + { + "batch_loss": 1.6604290008544922, + "innovation_rms": 0.0019775287236426142, + "instruction_rms": 0.0019775287061458563, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.02967562898993492, + "max_absolute_post_projection_soma_slope": 8.482015978472646e-09, + "max_absolute_pre_projection_soma_slope": 0.02967562898993492, + "max_positive_post_projection_soma_slope": 5.961901639039979e-09, + "min_post_projection_soma_slope": -8.482015978472646e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.819185965282357e-08, + "pre_projection_traffic_rms_ratio": 0.005291200951930861 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.354832649230957, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.48795318603516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03579501807689667, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8694826364517212, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18434613943099976, + "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": 0.7770912647247314, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0770423635840416, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6855415105819702, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0859965831041336, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03924165666103363, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.761473536491394, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1844303160905838, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1472022533416748, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02061918463525888, + "step": 307, + "teaching_rms": 0.0019775287236426142, + "traffic_rms": 0.02052215717266241 + }, + { + "batch_loss": 1.5826926231384277, + "innovation_rms": 0.00204636860953538, + "instruction_rms": 0.0020463686216126422, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 3.8656531309300135e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.8656531309300135e-09, + "min_post_projection_soma_slope": -3.4195328790787016e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8419633534102807e-08, + "pre_projection_traffic_rms_ratio": 0.0054030351183075205 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.340503692626953, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.30013275146484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.047123417258262634, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8624744415283203, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12381723523139954, + "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": 0.7802709937095642, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07395422458648682, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.683440923690796, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0997040718793869, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09590316563844681, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7502942681312561, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12377818673849106, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1470096111297607, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020642155866979792, + "step": 308, + "teaching_rms": 0.00204636860953538, + "traffic_rms": 0.020535980175261325 + }, + { + "batch_loss": 1.5716532468795776, + "innovation_rms": 0.00198987417363395, + "instruction_rms": 0.001989874186054097, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 9.127674616138393e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 6.305652888727309e-09, + "min_post_projection_soma_slope": -9.127674616138393e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8335563993731605e-08, + "pre_projection_traffic_rms_ratio": 0.005246028664149664 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.324873924255371, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.38068389892578, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05531531572341919, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8638448715209961, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1447426974773407, + "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": 0.7821921110153198, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06367504596710205, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.6934008598327637, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.099599689245224, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12189662456512451, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7511231303215027, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14477717876434326, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.14696204662323, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020670826370783883, + "step": 309, + "teaching_rms": 0.00198987417363395, + "traffic_rms": 0.020573024059838966 + }, + { + "batch_loss": 1.5637991428375244, + "innovation_rms": 0.002089851461109632, + "instruction_rms": 0.002089851484761581, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029334167018532753, + "max_absolute_post_projection_soma_slope": 8.550579799759817e-09, + "max_absolute_pre_projection_soma_slope": 0.029334167018532753, + "max_positive_post_projection_soma_slope": 8.550579799759817e-09, + "min_post_projection_soma_slope": -5.237855926765178e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8208402256930037e-08, + "pre_projection_traffic_rms_ratio": 0.004865661932326606 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.298943519592285, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.96175384521484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09693548083305359, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8649680018424988, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1750093698501587, + "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": 0.784712016582489, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09673524647951126, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7019519805908203, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08551101386547089, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14876452088356018, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.750700056552887, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17504385113716125, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1470268964767456, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02071744122418577, + "step": 310, + "teaching_rms": 0.002089851461109632, + "traffic_rms": 0.02060849130856164 + }, + { + "batch_loss": 1.6693947315216064, + "innovation_rms": 0.002175315041734705, + "instruction_rms": 0.002175315059912875, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 9.883889262596313e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 9.883889262596313e-09, + "min_post_projection_soma_slope": -7.464993068140302e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.830653146408722e-08, + "pre_projection_traffic_rms_ratio": 0.005513289549394805 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.28345012664795, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.16032409667969, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08920779824256897, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8705320954322815, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14248324930667877, + "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": 0.7834699749946594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11741125583648682, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7106505632400513, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0897514745593071, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1136142835021019, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7577459812164307, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14251771569252014, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1469508409500122, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02076911270455786, + "step": 311, + "teaching_rms": 0.002175315041734705, + "traffic_rms": 0.020655136749797742 + }, + { + "batch_loss": 1.700202226638794, + "innovation_rms": 0.0022166261812315018, + "instruction_rms": 0.0022166261812315018, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.863375118764225e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.863375118764225e-09, + "min_post_projection_soma_slope": -4.653530893961033e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8400537223797604e-08, + "pre_projection_traffic_rms_ratio": 0.005434955704945126 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.278926849365234, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.39827728271484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.056644756346940994, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8783221244812012, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10581167042255402, + "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": 0.7795080542564392, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11081204563379288, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.719396710395813, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08932565152645111, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1128683090209961, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7683187127113342, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11348097026348114, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.146704077720642, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020775540761784256, + "step": 312, + "teaching_rms": 0.0022166261812315018, + "traffic_rms": 0.02065846131701144 + }, + { + "batch_loss": 1.5015679597854614, + "innovation_rms": 0.0019258252143113624, + "instruction_rms": 0.0019258252168780055, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.7641721678814974e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 4.7641721678814974e-09, + "min_post_projection_soma_slope": -3.4244496127655566e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8421259536521396e-08, + "pre_projection_traffic_rms_ratio": 0.005084176629895635 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.268233299255371, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.89990997314453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.055896177887916565, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8859243392944336, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13378646969795227, + "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": 0.7752112150192261, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12341180443763733, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7295883893966675, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10191649198532104, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09813864529132843, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7770556807518005, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1341417133808136, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1465636491775513, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02075601957740691, + "step": 313, + "teaching_rms": 0.0019258252143113624, + "traffic_rms": 0.020665705091447776 + }, + { + "batch_loss": 1.7202198505401611, + "innovation_rms": 0.002308065283776482, + "instruction_rms": 0.002308065275210163, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 6.740368707625066e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 6.24164497864399e-09, + "min_post_projection_soma_slope": -6.740368707625066e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8198743553132523e-08, + "pre_projection_traffic_rms_ratio": 0.005295724264397562 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.248705863952637, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.29132080078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.03907446190714836, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8974300026893616, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.23429268598556519, + "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": 0.773428201675415, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11296841502189636, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7386119365692139, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10944047570228577, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08560071885585785, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7913628220558167, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2346053123474121, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1465246677398682, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020843685309979095, + "step": 314, + "teaching_rms": 0.002308065283776482, + "traffic_rms": 0.020715461093740797 + }, + { + "batch_loss": 1.6617366075515747, + "innovation_rms": 0.0023213541284072205, + "instruction_rms": 0.00232135412414858, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 3.906729162395095e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 3.906729162395095e-09, + "min_post_projection_soma_slope": -3.2591802590076213e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8380490433539306e-08, + "pre_projection_traffic_rms_ratio": 0.005341545420355491 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.238183975219727, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.27378845214844, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04247220605611801, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.909980058670044, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2037753164768219, + "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": 0.7728139758110046, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11413542926311493, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7458927631378174, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07785538583993912, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08477754145860672, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8064756989479065, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20408791303634644, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1466807126998901, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020867624263437957, + "step": 315, + "teaching_rms": 0.0023213541284072205, + "traffic_rms": 0.020738831108712406 + }, + { + "batch_loss": 1.4960441589355469, + "innovation_rms": 0.002329595420487659, + "instruction_rms": 0.002329595414122297, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.141596843254547e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 2.7394637669431177e-09, + "min_post_projection_soma_slope": -4.141596843254547e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.841670680978855e-08, + "pre_projection_traffic_rms_ratio": 0.005215963806457253 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.228447914123535, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.34142303466797, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05444063991308212, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9189862012863159, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14943455159664154, + "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": 0.7685889601707458, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12549473345279694, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.752576231956482, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10667209327220917, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08233058452606201, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8164389729499817, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1497896909713745, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.146983027458191, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020853006046932087, + "step": 316, + "teaching_rms": 0.002329595420487659, + "traffic_rms": 0.02072271673805619 + }, + { + "batch_loss": 1.6032092571258545, + "innovation_rms": 0.0024732969710687075, + "instruction_rms": 0.0024732969710687075, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 8.08861333467803e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 3.635345580121907e-09, + "min_post_projection_soma_slope": -8.08861333467803e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.841349227042287e-08, + "pre_projection_traffic_rms_ratio": 0.005222341340931025 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.217741012573242, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.34590911865234, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05886431783437729, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9212758541107178, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12910614907741547, + "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": 0.7655954360961914, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11231402307748795, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7610743045806885, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08995486795902252, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08717066794633865, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8193292617797852, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12910063564777374, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1470853090286255, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0208874728603497, + "step": 317, + "teaching_rms": 0.0024732969710687075, + "traffic_rms": 0.020741505358362115 + }, + { + "batch_loss": 1.5178669691085815, + "innovation_rms": 0.0024897156634650047, + "instruction_rms": 0.002489715665450334, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 5.421196380694937e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 5.421196380694937e-09, + "min_post_projection_soma_slope": -4.449366208802985e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.830415876549817e-08, + "pre_projection_traffic_rms_ratio": 0.005429340119192554 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.206642150878906, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.92925262451172, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05576472729444504, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9213554263114929, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16938060522079468, + "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": 0.7636500597000122, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09430874139070511, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7726609706878662, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1183089017868042, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11000999808311462, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.817330539226532, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16905269026756287, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1470847129821777, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02088776645272988, + "step": 318, + "teaching_rms": 0.0024897156634650047, + "traffic_rms": 0.020743767075263507 + }, + { + "batch_loss": 1.7414785623550415, + "innovation_rms": 0.0026121090799569578, + "instruction_rms": 0.0026121090799569578, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 5.249125578643543e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 5.249125578643543e-09, + "min_post_projection_soma_slope": -4.69011851578216e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8127152808054986e-08, + "pre_projection_traffic_rms_ratio": 0.005353133950709189 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.201079368591309, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.87286376953125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05735982581973076, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9144454598426819, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22006836533546448, + "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": 0.760711133480072, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09206761419773102, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7818416357040405, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12958915531635284, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1005907878279686, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8062875866889954, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.21974048018455505, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1468653678894043, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020871197729581688, + "step": 319, + "teaching_rms": 0.0026121090799569578, + "traffic_rms": 0.020710532595956125 + }, + { + "batch_loss": 1.4756861925125122, + "innovation_rms": 0.00246585165950407, + "instruction_rms": 0.0024658516615086133, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 3.509805779344788e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.509805779344788e-09, + "min_post_projection_soma_slope": -3.262889292088289e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8364257832956935e-08, + "pre_projection_traffic_rms_ratio": 0.005499057175366972 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.203340530395508, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.56400299072266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04437735304236412, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9013490080833435, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22734667360782623, + "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": 0.7599300742149353, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07961220294237137, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.790557861328125, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11144526302814484, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07688627392053604, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7880907654762268, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.22703425586223602, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1466909646987915, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020814507412777825, + "step": 320, + "teaching_rms": 0.00246585165950407, + "traffic_rms": 0.020666541456001712 + }, + { + "batch_loss": 1.7730739116668701, + "innovation_rms": 0.0024573750112088645, + "instruction_rms": 0.002457375017243238, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675614088773727, + "max_absolute_post_projection_soma_slope": 8.422372133054523e-09, + "max_absolute_pre_projection_soma_slope": 0.029675614088773727, + "max_positive_post_projection_soma_slope": 8.422372133054523e-09, + "min_post_projection_soma_slope": -2.3775201807296753e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8261197764167768e-08, + "pre_projection_traffic_rms_ratio": 0.005603130738789109 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.20693302154541, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.7290267944336, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05896080285310745, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8876888155937195, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20146559178829193, + "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": 0.7643795609474182, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07752955704927444, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.801173210144043, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1061539575457573, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11096510291099548, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7715813517570496, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2011532038450241, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1465731859207153, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020837788660126488, + "step": 321, + "teaching_rms": 0.0024573750112088645, + "traffic_rms": 0.020693994614313843 + }, + { + "batch_loss": 1.6309581995010376, + "innovation_rms": 0.002568912753391909, + "instruction_rms": 0.002568912761088404, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 6.938826846436541e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 6.211585024118449e-09, + "min_post_projection_soma_slope": -6.938826846436541e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.837753721437215e-08, + "pre_projection_traffic_rms_ratio": 0.005480735004297561 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.21888542175293, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.68260192871094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06447170674800873, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8751929998397827, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1655026078224182, + "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": 0.7706655859947205, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10962739586830139, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.811048984527588, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09875732660293579, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17827208340168, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7550227046012878, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1655866950750351, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1461327075958252, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02080391387268778, + "step": 322, + "teaching_rms": 0.002568912753391909, + "traffic_rms": 0.020640939648875683 + }, + { + "batch_loss": 1.6158759593963623, + "innovation_rms": 0.002413744710444263, + "instruction_rms": 0.0024137447145398962, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 1.1445755454531081e-08, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 1.1445755454531081e-08, + "min_post_projection_soma_slope": -5.4024642537342515e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8326125782071498e-08, + "pre_projection_traffic_rms_ratio": 0.005343984017839667 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.217461585998535, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.38763427734375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06031085550785065, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.869284451007843, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1510208249092102, + "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": 0.7824082374572754, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1470279097557068, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8213235139846802, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13453979790210724, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1968972235918045, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7456408739089966, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15096081793308258, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1456547975540161, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02082664596479251, + "step": 323, + "teaching_rms": 0.002413744710444263, + "traffic_rms": 0.02068041963130928 + }, + { + "batch_loss": 1.64281165599823, + "innovation_rms": 0.002296576122433809, + "instruction_rms": 0.0022965761138246344, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 3.1363251995486507e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 3.1363251995486507e-09, + "min_post_projection_soma_slope": -2.612074556651578e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8282307579464414e-08, + "pre_projection_traffic_rms_ratio": 0.005348012911261508 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.22116470336914, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.73542785644531, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05431370809674263, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8713459968566895, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1486043930053711, + "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": 0.7923980951309204, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13025197386741638, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.829663872718811, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.150587260723114, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15652567148208618, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7473543882369995, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14854440093040466, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1452059745788574, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02076560258342055, + "step": 324, + "teaching_rms": 0.002296576122433809, + "traffic_rms": 0.020637658616558103 + }, + { + "batch_loss": 1.8121098279953003, + "innovation_rms": 0.0023931800673477346, + "instruction_rms": 0.002393180079740215, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 7.152333392212995e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.594221003628718e-09, + "min_post_projection_soma_slope": -7.152333392212995e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.835509855475339e-08, + "pre_projection_traffic_rms_ratio": 0.005587489513942384 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.237263679504395, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.48094177246094, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06189221516251564, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8726896047592163, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13859157264232635, + "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": 0.8041454553604126, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11747359484434128, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8365603685379028, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14577767252922058, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13470445573329926, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7491205334663391, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13853158056735992, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1447103023529053, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020847838375371183, + "step": 325, + "teaching_rms": 0.0023931800673477346, + "traffic_rms": 0.02071281550413558 + }, + { + "batch_loss": 1.7061867713928223, + "innovation_rms": 0.002279434048068192, + "instruction_rms": 0.0022794340437312334, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 6.560763932128566e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.311135448631376e-09, + "min_post_projection_soma_slope": -6.560763932128566e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8292774882354163e-08, + "pre_projection_traffic_rms_ratio": 0.0056565113200449805 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.261592864990234, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.3585205078125, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07480628788471222, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8771900534629822, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1829785704612732, + "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": 0.8141366839408875, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09991218894720078, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8404760360717773, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11806442588567734, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09933669865131378, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7534600496292114, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18301300704479218, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1441712379455566, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020846255366756047, + "step": 326, + "teaching_rms": 0.002279434048068192, + "traffic_rms": 0.020724688314197992 + }, + { + "batch_loss": 1.670188069343567, + "innovation_rms": 0.002021965032148645, + "instruction_rms": 0.0020219650419270653, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.527005437182652e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 2.245325481098348e-09, + "min_post_projection_soma_slope": -4.527005437182652e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8171696689547756e-08, + "pre_projection_traffic_rms_ratio": 0.005580362643682818 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.282196044921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.10940551757812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05896750092506409, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8833592534065247, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19852440059185028, + "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": 0.8231530785560608, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10056339204311371, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8395042419433594, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15300852060317993, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07173924893140793, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7602854371070862, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.19883660972118378, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1436835527420044, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02087306423419406, + "step": 327, + "teaching_rms": 0.002021965032148645, + "traffic_rms": 0.02077416852439293 + }, + { + "batch_loss": 1.7211887836456299, + "innovation_rms": 0.002103009526422978, + "instruction_rms": 0.0021030095240725816, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 4.306452083824297e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 1.7159377252440322e-09, + "min_post_projection_soma_slope": -4.306452083824297e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.840244274718201e-08, + "pre_projection_traffic_rms_ratio": 0.005646096971902344 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.310907363891602, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.46536254882812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05150527507066727, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8885030746459961, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2106112688779831, + "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": 0.8296801447868347, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09627579897642136, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8361873626708984, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15462040901184082, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07141222804784775, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7662926912307739, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.2109234482049942, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.143259882926941, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020836158103972782, + "step": 328, + "teaching_rms": 0.002103009526422978, + "traffic_rms": 0.020731764289209184 + }, + { + "batch_loss": 1.577652096748352, + "innovation_rms": 0.0020106121488453342, + "instruction_rms": 0.002010612142699313, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675615951418877, + "max_absolute_post_projection_soma_slope": 8.396579431746432e-09, + "max_absolute_pre_projection_soma_slope": 0.029675615951418877, + "max_positive_post_projection_soma_slope": 5.483661968952447e-09, + "min_post_projection_soma_slope": -8.396579431746432e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8466242771962124e-08, + "pre_projection_traffic_rms_ratio": 0.005743344855204308 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.334794998168945, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.70755004882812, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.051930177956819534, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.892160952091217, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20326924324035645, + "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": 0.8316868543624878, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09723205864429474, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.831125020980835, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1107320636510849, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08398493379354477, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7718314528465271, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.20358139276504517, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1426767110824585, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02083244398491279, + "step": 329, + "teaching_rms": 0.0020106121488453342, + "traffic_rms": 0.020739957927792298 + }, + { + "batch_loss": 1.6636189222335815, + "innovation_rms": 0.0019710714743695107, + "instruction_rms": 0.001971071473115648, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 5.85266812791474e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 1.6024069848796785e-09, + "min_post_projection_soma_slope": -5.85266812791474e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8281108398109226e-08, + "pre_projection_traffic_rms_ratio": 0.005607475991095935 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.357684135437012, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.9081039428711, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04892382025718689, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8943712115287781, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14464125037193298, + "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": 0.831057608127594, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08873943984508514, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8265793323516846, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11485391855239868, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08431770652532578, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7763072848320007, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14495337009429932, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1421504020690918, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02084408969138586, + "step": 330, + "teaching_rms": 0.0019710714743695107, + "traffic_rms": 0.02075167643507877 + }, + { + "batch_loss": 1.654758095741272, + "innovation_rms": 0.002050576520617507, + "instruction_rms": 0.002050576525438499, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.121238017518181e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 3.972955742170825e-09, + "min_post_projection_soma_slope": -4.121238017518181e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8340224571491134e-08, + "pre_projection_traffic_rms_ratio": 0.005619354001815019 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.383450508117676, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.44761657714844, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05760921165347099, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8939089179039001, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09825068712234497, + "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": 0.8325091004371643, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09501239657402039, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8196316957473755, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1326684206724167, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1152290403842926, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7782440781593323, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09807185083627701, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.141737461090088, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020825669151806905, + "step": 331, + "teaching_rms": 0.002050576520617507, + "traffic_rms": 0.02072721430774075 + }, + { + "batch_loss": 1.565643072128296, + "innovation_rms": 0.0020391530224953486, + "instruction_rms": 0.0020391530237073483, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.9441510319734334e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 3.161754635883085e-09, + "min_post_projection_soma_slope": -6.9441510319734334e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.846362804736606e-08, + "pre_projection_traffic_rms_ratio": 0.005833266617383342 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.416240692138672, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.47359466552734, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.052322302013635635, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8959020972251892, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0951157733798027, + "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": 0.8328462839126587, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07262267172336578, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8137413263320923, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11693914234638214, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11933595687150955, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7834995985031128, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09532251954078674, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1414425373077393, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020777728020352142, + "step": 332, + "teaching_rms": 0.0020391530224953486, + "traffic_rms": 0.020676777049259198 + }, + { + "batch_loss": 1.5240089893341064, + "innovation_rms": 0.001905803283741159, + "instruction_rms": 0.0019058032746635321, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 4.2898613550335085e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 4.0879801765925095e-09, + "min_post_projection_soma_slope": -4.2898613550335085e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8376164195578473e-08, + "pre_projection_traffic_rms_ratio": 0.005151040338235162 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.446258544921875, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 66.81302642822266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.042311426252126694, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8986254334449768, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10556004196405411, + "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": 0.8288484811782837, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10383421182632446, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8077726364135742, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11419379711151123, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11312800645828247, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7873747944831848, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10567821562290192, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1411579847335815, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020716954092170694, + "step": 333, + "teaching_rms": 0.001905803283741159, + "traffic_rms": 0.020630074872507762 + }, + { + "batch_loss": 1.6060198545455933, + "innovation_rms": 0.0019833725895830353, + "instruction_rms": 0.0019833726020438962, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 5.335984543108907e-09, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 4.236770045906724e-09, + "min_post_projection_soma_slope": -5.335984543108907e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8242987940790322e-08, + "pre_projection_traffic_rms_ratio": 0.005492005630542934 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.473873138427734, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.05358123779297, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04978710412979126, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.900098443031311, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12036028504371643, + "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": 0.8231005668640137, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11725781857967377, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.802610158920288, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12964442372322083, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10835196822881699, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7901034355163574, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13909417390823364, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1409447193145752, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020730152548131302, + "step": 334, + "teaching_rms": 0.0019833725895830353, + "traffic_rms": 0.02063657620223169 + }, + { + "batch_loss": 1.6082289218902588, + "innovation_rms": 0.0020263969072050834, + "instruction_rms": 0.0020263969047658245, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 5.782990530889265e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 2.9687363678476686e-09, + "min_post_projection_soma_slope": -5.782990530889265e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.831775961589007e-08, + "pre_projection_traffic_rms_ratio": 0.005468817134326765 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.506026268005371, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.41080474853516, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04917663335800171, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8990182280540466, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13064990937709808, + "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": 0.8171120285987854, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1353769302368164, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7992503643035889, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12993904948234558, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10488477349281311, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7883538603782654, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16458815336227417, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.140674114227295, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020700929977203325, + "step": 335, + "teaching_rms": 0.0020263969072050834, + "traffic_rms": 0.020600956446565816 + }, + { + "batch_loss": 1.6233224868774414, + "innovation_rms": 0.002071761689101275, + "instruction_rms": 0.002071761679557888, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 7.172713090142224e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 7.172713090142224e-09, + "min_post_projection_soma_slope": -2.5928530433816377e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.833515280542101e-08, + "pre_projection_traffic_rms_ratio": 0.0053765882492689855 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.536213874816895, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.86077880859375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05190880596637726, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8975446224212646, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.16424129903316498, + "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": 0.8077492117881775, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1671341061592102, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.795335054397583, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13922564685344696, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12348003685474396, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7832770943641663, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18629911541938782, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.140537977218628, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02059634910492424, + "step": 336, + "teaching_rms": 0.002071761689101275, + "traffic_rms": 0.02049181905108071 + }, + { + "batch_loss": 1.6417304277420044, + "innovation_rms": 0.002245169326247965, + "instruction_rms": 0.0022451693174416697, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 5.065271313497988e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 2.853510761013922e-09, + "min_post_projection_soma_slope": -5.065271313497988e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8399457346182698e-08, + "pre_projection_traffic_rms_ratio": 0.005614662682770134 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.559038162231445, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.24198150634766, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05346090719103813, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8954591155052185, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1536441594362259, + "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": 0.7968803644180298, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17479239404201508, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7939906120300293, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09290896356105804, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1229417473077774, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7765127420425415, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17930461466312408, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1404170989990234, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02061953670765039, + "step": 337, + "teaching_rms": 0.002245169326247965, + "traffic_rms": 0.02049577791343634 + }, + { + "batch_loss": 1.6758549213409424, + "innovation_rms": 0.0023587786929827105, + "instruction_rms": 0.002358778684600566, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 7.184273620453041e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 5.560254479064497e-09, + "min_post_projection_soma_slope": -7.184273620453041e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8493775530042263e-08, + "pre_projection_traffic_rms_ratio": 0.005611874367677974 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.581141471862793, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.4612808227539, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04751831293106079, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9000271558761597, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11556422710418701, + "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": 0.7850567102432251, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15208210051059723, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7919471263885498, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10765408724546432, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1312476247549057, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7755382061004639, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1255892813205719, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1401989459991455, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020657679598784593, + "step": 338, + "teaching_rms": 0.0023587786929827105, + "traffic_rms": 0.020519991422675868 + }, + { + "batch_loss": 1.4962210655212402, + "innovation_rms": 0.0023184482129359994, + "instruction_rms": 0.0023184482171999775, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.7043991752436796e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.674157644795173e-09, + "min_post_projection_soma_slope": -5.7043991752436796e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8550186866023347e-08, + "pre_projection_traffic_rms_ratio": 0.005467034166177265 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.59189510345459, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.8514175415039, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05855795741081238, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9056145548820496, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11273697018623352, + "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": 0.7743129730224609, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12254201620817184, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.794769048690796, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14037226140499115, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.139289990067482, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7760080695152283, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11298053711652756, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1399606466293335, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020646142910029544, + "step": 339, + "teaching_rms": 0.0023184482129359994, + "traffic_rms": 0.020510417296842866 + }, + { + "batch_loss": 1.7414896488189697, + "innovation_rms": 0.0022675158976378155, + "instruction_rms": 0.0022675158932780612, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 7.56614504382469e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 3.893080968708773e-09, + "min_post_projection_soma_slope": -7.56614504382469e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.849246729583008e-08, + "pre_projection_traffic_rms_ratio": 0.005317392740901602 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.592077255249023, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 67.66923522949219, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07073421776294708, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9103021621704102, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11084333062171936, + "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": 0.7657625675201416, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13047219812870026, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.7964028120040894, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11122242361307144, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11317578703165054, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7755709290504456, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1108274981379509, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1397583484649658, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02066212699089077, + "step": 340, + "teaching_rms": 0.0022675158976378155, + "traffic_rms": 0.020537989193544725 + }, + { + "batch_loss": 1.6083874702453613, + "innovation_rms": 0.0020961225080178792, + "instruction_rms": 0.0020961225080178792, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 6.378048755806276e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 2.9368170118004855e-09, + "min_post_projection_soma_slope": -6.378048755806276e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.850259313281778e-08, + "pre_projection_traffic_rms_ratio": 0.005409867909874043 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.593189239501953, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 68.0111083984375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06428439170122147, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9087437987327576, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10482772439718246, + "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": 0.7573280930519104, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10700228065252304, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8006622791290283, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12488614022731781, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11488132178783417, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7691763043403625, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10499914735555649, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1397961378097534, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020684818138240506, + "step": 341, + "teaching_rms": 0.0020961225080178792, + "traffic_rms": 0.020576416304035224 + }, + { + "batch_loss": 1.6920591592788696, + "innovation_rms": 0.002518571656405111, + "instruction_rms": 0.002518571648554779, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 5.242512646219666e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 5.242512646219666e-09, + "min_post_projection_soma_slope": -3.6062386410407044e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8418888086018527e-08, + "pre_projection_traffic_rms_ratio": 0.005506673089071579 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.590027809143066, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 69.18265533447266, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04640022665262222, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9041467905044556, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15251781046390533, + "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": 0.7547343969345093, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11912374198436737, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8051131963729858, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13562315702438354, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08821532875299454, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7603894472122192, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15268921852111816, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.139865517616272, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02076708868861987, + "step": 342, + "teaching_rms": 0.002518571656405111, + "traffic_rms": 0.02061118616841572 + }, + { + "batch_loss": 1.6282289028167725, + "innovation_rms": 0.002278121484188536, + "instruction_rms": 0.002278121486358265, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 7.55524620643655e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 7.55524620643655e-09, + "min_post_projection_soma_slope": -6.651152961723028e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8410465597225448e-08, + "pre_projection_traffic_rms_ratio": 0.005362893765621739 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.587300300598145, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 70.17538452148438, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.050838690251111984, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8953422904014587, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1528746485710144, + "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": 0.7562367916107178, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.0981181412935257, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8122875690460205, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08692186325788498, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07618216425180435, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7499441504478455, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.15288013219833374, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1399242877960205, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02079498327057038, + "step": 343, + "teaching_rms": 0.002278121484188536, + "traffic_rms": 0.020667668570122876 + }, + { + "batch_loss": 1.6986024379730225, + "innovation_rms": 0.0024448981682048997, + "instruction_rms": 0.002444898162139732, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 1.2217705958050828e-08, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 1.2217705958050828e-08, + "min_post_projection_soma_slope": -2.544844335261587e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8416260079992086e-08, + "pre_projection_traffic_rms_ratio": 0.005277276003608574 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.592207908630371, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 71.49940490722656, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06249324232339859, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8928907513618469, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17511975765228271, + "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": 0.7590948939323425, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10160116106271744, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8212922811508179, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11590094864368439, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10823139548301697, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7462249398231506, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17512525618076324, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1398940086364746, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020831023749722802, + "step": 344, + "teaching_rms": 0.0024448981682048997, + "traffic_rms": 0.020687289081125573 + }, + { + "batch_loss": 1.6574279069900513, + "innovation_rms": 0.002296474260730031, + "instruction_rms": 0.002296474273644365, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675625264644623, + "max_absolute_post_projection_soma_slope": 5.779855261067723e-09, + "max_absolute_pre_projection_soma_slope": 0.029675625264644623, + "max_positive_post_projection_soma_slope": 5.779855261067723e-09, + "min_post_projection_soma_slope": -3.781610136144309e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8446957197645284e-08, + "pre_projection_traffic_rms_ratio": 0.005468630270103626 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.603326797485352, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 72.74821472167969, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.054058998823165894, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8922417163848877, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14028429985046387, + "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": 0.7665756940841675, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10410335659980774, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.830665111541748, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09448199719190598, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13283883035182953, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7447041273117065, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14018218219280243, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1399164199829102, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020882468364178394, + "step": 345, + "teaching_rms": 0.002296474260730031, + "traffic_rms": 0.020755558068753815 + }, + { + "batch_loss": 1.5546998977661133, + "innovation_rms": 0.002152194966840881, + "instruction_rms": 0.002152194966840881, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 1.3250149422106006e-08, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 1.3250149422106006e-08, + "min_post_projection_soma_slope": -6.487542059119278e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.837429300035668e-08, + "pre_projection_traffic_rms_ratio": 0.0052705725082258905 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.61383056640625, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 73.7415771484375, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04633348807692528, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8962675929069519, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17853128910064697, + "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": 0.774816632270813, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11034324765205383, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8378596305847168, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09057539701461792, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.08355458080768585, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7481419444084167, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17842918634414673, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.139962911605835, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02087984695599905, + "step": 346, + "teaching_rms": 0.002152194966840881, + "traffic_rms": 0.020770459341177765 + }, + { + "batch_loss": 1.5558606386184692, + "innovation_rms": 0.00219413196676135, + "instruction_rms": 0.0021941319622557813, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675621539354324, + "max_absolute_post_projection_soma_slope": 6.190336687694753e-09, + "max_absolute_pre_projection_soma_slope": 0.029675621539354324, + "max_positive_post_projection_soma_slope": 6.190336687694753e-09, + "min_post_projection_soma_slope": -4.9036534832680445e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.845192345855954e-08, + "pre_projection_traffic_rms_ratio": 0.005181903844108197 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.637165069580078, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 74.70646667480469, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04833662509918213, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8999315500259399, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12794683873653412, + "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": 0.786077618598938, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11261013150215149, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8463059663772583, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1014183834195137, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.05360158905386925, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7518594861030579, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12796318531036377, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1400556564331055, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.02091330883327367, + "step": 347, + "teaching_rms": 0.00219413196676135, + "traffic_rms": 0.02079781971462749 + }, + { + "batch_loss": 1.653876781463623, + "innovation_rms": 0.002232243058975338, + "instruction_rms": 0.002232243065618305, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 6.060568047416837e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 2.897476480967498e-09, + "min_post_projection_soma_slope": -6.060568047416837e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8453546725926594e-08, + "pre_projection_traffic_rms_ratio": 0.005305842450757391 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.678326606750488, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.32044982910156, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04504476487636566, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.897575855255127, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14676842093467712, + "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": 0.7991215586662292, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13043923676013947, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8544450998306274, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09702155739068985, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.07003191113471985, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.749887228012085, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1466989666223526, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1397770643234253, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020944478151697105, + "step": 348, + "teaching_rms": 0.002232243058975338, + "traffic_rms": 0.020824429683272996 + }, + { + "batch_loss": 1.4300158023834229, + "innovation_rms": 0.0021857663103557095, + "instruction_rms": 0.0021857663103557095, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675617814064026, + "max_absolute_post_projection_soma_slope": 9.411463608444137e-09, + "max_absolute_pre_projection_soma_slope": 0.029675617814064026, + "max_positive_post_projection_soma_slope": 4.881172355197805e-09, + "min_post_projection_soma_slope": -9.411463608444137e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8393990235628912e-08, + "pre_projection_traffic_rms_ratio": 0.005368833563367806 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.723895072937012, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 75.77944946289062, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04600130021572113, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8952004909515381, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18304866552352905, + "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": 0.8140569925308228, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14935415983200073, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8624218702316284, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.12674923241138458, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.10236579179763794, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7486996650695801, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18297922611236572, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1393312215805054, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.020930864527129896, + "step": 349, + "teaching_rms": 0.0021857663103557095, + "traffic_rms": 0.020813198414491044 + }, + { + "batch_loss": 1.6277883052825928, + "innovation_rms": 0.0023129320690190506, + "instruction_rms": 0.0023129320647449036, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675627127289772, + "max_absolute_post_projection_soma_slope": 1.089524470643255e-08, + "max_absolute_pre_projection_soma_slope": 0.029675627127289772, + "max_positive_post_projection_soma_slope": 3.326879882692424e-09, + "min_post_projection_soma_slope": -1.089524470643255e-08, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.8472678541318977e-08, + "pre_projection_traffic_rms_ratio": 0.005208528878880821 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.778693199157715, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.37044525146484, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.054673705250024796, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8953872919082642, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17475822567939758, + "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": 0.8274642825126648, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13407295942306519, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8677862882614136, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.14972573518753052, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.114149309694767, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7505752444267273, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17468880116939545, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1388617753982544, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.0209720920554896, + "step": 350, + "teaching_rms": 0.0023129320690190506, + "traffic_rms": 0.020843079931667632 + }, + { + "batch_loss": 1.5544573068618774, + "innovation_rms": 0.0021565595067114915, + "instruction_rms": 0.002156559502127425, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675623401999474, + "max_absolute_post_projection_soma_slope": 4.500781525251796e-09, + "max_absolute_pre_projection_soma_slope": 0.029675623401999474, + "max_positive_post_projection_soma_slope": 4.014019339138031e-09, + "min_post_projection_soma_slope": -4.500781525251796e-09, + "observations": 128, + "post_projection_traffic_rms_ratio": 2.824632569751395e-08, + "pre_projection_traffic_rms_ratio": 0.005365862820073683 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.83841323852539, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.54297637939453, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.04903523996472359, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.8995904922485352, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.17149342596530914, + "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": 0.8412848711013794, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13820606470108032, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8695439100265503, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1524190455675125, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1354191154241562, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7572043538093567, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.171424001455307, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.138375997543335, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021013636051452497, + "step": 351, + "teaching_rms": 0.0021565595067114915, + "traffic_rms": 0.020902863413321114 + }, + { + "batch_loss": 1.5239646434783936, + "innovation_rms": 0.0020874714535660064, + "instruction_rms": 0.0020874714619851763, + "neutral_projection": { + "instruction_observations": 0, + "max_absolute_correction_slope": 0.029675619676709175, + "max_absolute_post_projection_soma_slope": 4.062326919296311e-09, + "max_absolute_pre_projection_soma_slope": 0.029675619676709175, + "max_positive_post_projection_soma_slope": 4.062326919296311e-09, + "min_post_projection_soma_slope": -2.5563160477304336e-09, + "observations": 72, + "post_projection_traffic_rms_ratio": 2.8075511286737367e-08, + "pre_projection_traffic_rms_ratio": 0.005319258034274626 + }, + "parameter_state": { + "bn_running_mean": { + "all_finite": true, + "max_abs_over_finite_tensors": 8.923418998718262, + "nonfinite_indices": [] + }, + "bn_running_var": { + "all_finite": true, + "max_abs_over_finite_tensors": 76.2424087524414, + "nonfinite_indices": [] + }, + "feedback_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.06286109238862991, + "nonfinite_indices": [] + }, + "feedback_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.9061430096626282, + "nonfinite_indices": [] + }, + "feedback_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18172729015350342, + "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": 0.8473549485206604, + "nonfinite_indices": [] + }, + "forward_bn_bias_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.1251421868801117, + "nonfinite_indices": [] + }, + "forward_bn_scale": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.8703469038009644, + "nonfinite_indices": [] + }, + "forward_bn_scale_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.13730883598327637, + "nonfinite_indices": [] + }, + "forward_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.09399203211069107, + "nonfinite_indices": [] + }, + "forward_readout": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.7636339664459229, + "nonfinite_indices": [] + }, + "forward_readout_momentum": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.18189680576324463, + "nonfinite_indices": [] + }, + "forward_weight": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1377975940704346, + "nonfinite_indices": [] + }, + "predictor_bias": { + "all_finite": true, + "max_abs_over_finite_tensors": 1.1175870895385742e-08, + "nonfinite_indices": [] + }, + "predictor_slope": { + "all_finite": true, + "max_abs_over_finite_tensors": 0.11251720041036606, + "nonfinite_indices": [] + } + }, + "predictor_mse": null, + "predictor_updated": false, + "raw_apical_rms": 0.021049511736703372, + "step": 352, + "teaching_rms": 0.0020874714535660064, + "traffic_rms": 0.020948007579744698 + } + ], + "trajectory_metrics_finite": true, + "validation_evaluations": 0 +} diff --git a/results/kp_dynamic_projection_gate.json b/results/kp_dynamic_projection_gate.json new file mode 100644 index 0000000..ce6cccc --- /dev/null +++ b/results/kp_dynamic_projection_gate.json @@ -0,0 +1,42 @@ +{ + "checks": { + "final_32_mean_loss_at_most_2p5": true, + "forward_and_feedback_momentum_max_at_most_50": true, + "forward_and_feedback_weight_max_at_most_10": true, + "maximum_batch_loss_at_most_10": true, + "no_held_out_evaluations": true, + "paired_local_neutral_observations_only": true, + "post_projection_soma_slope_at_most_1e_minus_5": true, + "post_projection_traffic_ratio_at_most_1e_minus_5": true, + "record_trajectory_and_state_finite": true, + "slow_predictor_frozen_during_task": true, + "task_loader_state_restored": true, + "traffic_ratio_calibrated": true, + "used_instruction_rms_ratio_within_1e_minus_4": true, + "zero_margin_64_observation_slow_fit": true + }, + "metrics": { + "final_32_mean_loss": 1.629259716719389, + "maximum_batch_loss": 2.9280824661254883, + "maximum_correction_slope": 0.02967563085258007, + "maximum_feedback_momentum": 0.3636978566646576, + "maximum_feedback_weight": 0.8331910371780396, + "maximum_forward_momentum": 0.599919855594635, + "maximum_forward_weight": 1.1814264059066772, + "maximum_initial_traffic_ratio_error": 4.76837158203125e-07, + "maximum_post_projection_soma_slope": 1.5233538874781516e-08, + "maximum_post_projection_traffic_rms_ratio": 3.032565852668684e-08, + "maximum_pre_projection_traffic_rms_ratio": 0.005833266617383342, + "maximum_projection_observations": 128, + "maximum_used_instruction_rms_ratio_error": 2.1080730938649594e-08, + "minimum_projection_observations": 72 + }, + "protocol": "kp_dynamic_neutral_projection_training_prefix_v1", + "review_score_after": 5, + "review_score_before": 5, + "score_change_rule": "training-only stability evidence cannot change the paper score", + "source_commit": "8b28522305612d6300f485e14046ae38d49af426", + "status": "passed", + "test_evaluations": 0, + "validation_evaluations": 0 +} |