diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 15:08:36 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 15:08:36 -0500 |
| commit | 82d49f4b8b741669ab39100f4719719801064f20 (patch) | |
| tree | 7c7e7a1f58338508e89dab47b3bdf43eb7bf10d0 /results | |
| parent | 6fb38aabad3ce33a263414ecaf4eef0474c66045 (diff) | |
results: record residual mirror full failure
Diffstat (limited to 'results')
| -rw-r--r-- | results/residual_mirror_full/rrm.json | 3456 | ||||
| -rw-r--r-- | results/residual_mirror_full_gate.json | 35 |
2 files changed, 3491 insertions, 0 deletions
diff --git a/results/residual_mirror_full/rrm.json b/results/residual_mirror_full/rrm.json new file mode 100644 index 0000000..0784a73 --- /dev/null +++ b/results/residual_mirror_full/rrm.json @@ -0,0 +1,3456 @@ +{ + "architecture": { + "adaptive_apical_parameters": 0, + "adaptive_feedback_parameters": 267904, + "base_width": 16, + "blocks_per_stage": 3, + "bn_eps": 1e-05, + "bn_momentum": 0.1, + "depth": 20, + "family": "CIFAR 6n+2 ResNet, option-A shortcuts", + "fixed_feedback_parameters": 0, + "fixed_traffic_coefficients": 0, + "forward_parameters": 269722, + "hidden_shapes": [ + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 32, + 16, + 16 + ], + [ + 32, + 16, + 16 + ], + [ + 32, + 16, + 16 + ], + [ + 32, + 16, + 16 + ], + [ + 32, + 16, + 16 + ], + [ + 32, + 16, + 16 + ], + [ + 64, + 8, + 8 + ], + [ + 64, + 8, + 8 + ], + [ + 64, + 8, + 8 + ], + [ + 64, + 8, + 8 + ], + [ + 64, + 8, + 8 + ], + [ + 64, + 8, + 8 + ] + ], + "normalization": "batchnorm", + "predictor_parameters": 0, + "residual_scale": 1.0, + "vectorizer_mode": null, + "vectorizer_parameters": 0 + }, + "args": { + "a_scale": 1.0, + "a_warmup_steps": 0, + "alignment_probe": 32, + "apical_calibration_mode": "unit_targets", + "apical_seed": null, + "augment_train": 1, + "batch_size": 128, + "bn_eps": 1e-05, + "bn_momentum": 0.1, + "data_dir": "/home/yurenh2/sdrn/data", + "depth": 20, + "device": "cuda", + "epochs": 200, + "eta_A": 0.01, + "eta_P": 0.01, + "eval_every": 0, + "eval_split": "validation", + "learn_P": 0, + "loader_seed": 0, + "lr": 0.1, + "lr_gamma": 0.1, + "lr_milestones": "100,150", + "lr_schedule": "step", + "max_steps": 0, + "mirror_batch_size": 1, + "mirror_eta": 0.1, + "mirror_every": 16, + "mirror_noise_std": 1.0, + "mirror_seed": 3000, + "mirror_warmup_steps": 20, + "mode": "rrm", + "momentum": 0.9, + "normalization": "batchnorm", + "nuisance_scale": 0.0, + "out": "results/residual_mirror_full/rrm.json", + "output_lr": 0.1, + "pert_directions": 1, + "pert_every": 4, + "pert_sigma": 0.01, + "perturb_seed": 1000, + "predictor_warmup_steps": 0, + "residual_scale": null, + "seed": 0, + "split_seed": 2027, + "train_limit": 0, + "use_residual": 1, + "val_examples": 5000, + "vectorizer_mode": "spatial_template", + "warmup_epochs": 0, + "weight_decay": 0.0001, + "weight_scale": 1.0, + "width": 16 + }, + "calibration_metric_space": "local_parent_child_response_residual", + "counters": { + "apical_warmup_examples": 0, + "calibration_event_examples": 0, + "causal_scalar_observations": 0, + "logical_batch_loss_queries": 0, + "mirror_conv_examples": 4420, + "mirror_events": 4420, + "mirror_readout_examples": 4526080, + "ordinary_examples": 9000000, + "per_example_loss_terms": 0, + "perturbation_events": 0, + "perturbation_forward_examples": 0, + "predictor_warmup_examples": 0 + }, + "diagnostics": { + "early_third_mean": 0.8774353464444479, + "feedback_forward_cosine": [ + 1.0, + 1.0, + 1.0, + 1.0, + 1.0, + 1.0, + 1.0, + 1.0, + 0.9999998807907104, + 1.0, + 1.0, + 0.9999999403953552, + 1.0, + 0.9999999403953552, + 1.0, + 0.9999998807907104, + 0.9999998807907104, + 1.0, + 0.999961793422699 + ], + "feedback_forward_norm_ratio": [ + 1.000159502029419, + 1.000159740447998, + 1.0001599788665771, + 1.0001596212387085, + 1.0001596212387085, + 1.0001600980758667, + 1.000162959098816, + 1.0001602172851562, + 1.000161051750183, + 1.0001603364944458, + 1.0001589059829712, + 1.000158667564392, + 1.0001603364944458, + 1.0001589059829712, + 1.0001580715179443, + 1.0001592636108398, + 1.000157356262207, + 1.0001590251922607, + 1.0024465322494507 + ], + "innovation_negative_gradient_cosine": [ + 0.8722070455551147, + 0.8785926699638367, + 0.8789833784103394, + 0.877299964427948, + 0.8784912824630737, + 0.8790377378463745, + 0.8774803876876831, + 0.8692848086357117, + 0.8782165050506592, + 0.8673566579818726, + 0.8740392923355103, + 0.8405126929283142, + 0.8707835078239441, + 0.848376452922821, + 0.8770161867141724, + 0.8456900119781494, + 0.9081377387046814, + 0.8670427799224854, + 0.43746840953826904 + ], + "normalization_state": "training_batch_stats_without_running_update", + "raw_negative_gradient_cosine": [ + 0.8722070455551147, + 0.8785926699638367, + 0.8789833784103394, + 0.877299964427948, + 0.8784912824630737, + 0.8790377378463745, + 0.8774803876876831, + 0.8692848086357117, + 0.8782165050506592, + 0.8673566579818726, + 0.8740392923355103, + 0.8405126929283142, + 0.8707835078239441, + 0.848376452922821, + 0.8770161867141724, + 0.8456900119781494, + 0.9081377387046814, + 0.8670427799224854, + 0.43746840953826904 + ], + "teaching_negative_gradient_cosine": [ + 0.8722070455551147, + 0.8785926699638367, + 0.8789833784103394, + 0.877299964427948, + 0.8784912824630737, + 0.8790377378463745, + 0.8774803876876831, + 0.8692848086357117, + 0.8782165050506592, + 0.8673566579818726, + 0.8740392923355103, + 0.8405126929283142, + 0.8707835078239441, + 0.848376452922821, + 0.8770161867141724, + 0.8456900119781494, + 0.9081377387046814, + 0.8670427799224854, + 0.43746840953826904 + ], + "wall_s": 0.12804341316223145 + }, + "epochs": [ + { + "epoch": 1, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.21472547346632692, + "mirror_response_residual_rms": 0.2952180119297088, + "mirror_update_rms": 0.05842629268416273, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 352, + "train_examples": 45000, + "train_loss": 2.6106753037346735 + }, + { + "epoch": 2, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.15766319208710644, + "mirror_response_residual_rms": 0.2251705946689076, + "mirror_update_rms": 0.029419690545340785, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 704, + "train_examples": 45000, + "train_loss": 2.507496058061388 + }, + { + "epoch": 3, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14360009119075642, + "mirror_response_residual_rms": 0.21789556492020895, + "mirror_update_rms": 0.024176202518473574, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 1056, + "train_examples": 45000, + "train_loss": 2.3631224157333373 + }, + { + "epoch": 4, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.13956318963644174, + "mirror_response_residual_rms": 0.2275245609332117, + "mirror_update_rms": 0.023461374725354724, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 1408, + "train_examples": 45000, + "train_loss": 2.407446045939128 + }, + { + "epoch": 5, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14322111557358022, + "mirror_response_residual_rms": 0.2562621412969786, + "mirror_update_rms": 0.02647737383341734, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 1760, + "train_examples": 45000, + "train_loss": 3.116548478105333 + }, + { + "epoch": 6, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14567420421636174, + "mirror_response_residual_rms": 0.2909153129107003, + "mirror_update_rms": 0.030809725691105457, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 2112, + "train_examples": 45000, + "train_loss": 3.5446029083251953 + }, + { + "epoch": 7, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1501736402592817, + "mirror_response_residual_rms": 0.33786408134495804, + "mirror_update_rms": 0.03518108381263723, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 2464, + "train_examples": 45000, + "train_loss": 4.556180107116699 + }, + { + "epoch": 8, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1575769384844244, + "mirror_response_residual_rms": 0.40699926289597027, + "mirror_update_rms": 0.04229958157085732, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 2816, + "train_examples": 45000, + "train_loss": 6.37619294535319 + }, + { + "epoch": 9, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1527819485339171, + "mirror_response_residual_rms": 0.4540714081553837, + "mirror_update_rms": 0.04667154633073083, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 3168, + "train_examples": 45000, + "train_loss": 7.500557015991211 + }, + { + "epoch": 10, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1453256332382141, + "mirror_response_residual_rms": 0.4906493030655051, + "mirror_update_rms": 0.052956356645615044, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 3520, + "train_examples": 45000, + "train_loss": 8.279763238016765 + }, + { + "epoch": 11, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14547492011618718, + "mirror_response_residual_rms": 0.5564810488728783, + "mirror_update_rms": 0.05899267587832491, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 3872, + "train_examples": 45000, + "train_loss": 10.849900153944228 + }, + { + "epoch": 12, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14372628437196464, + "mirror_response_residual_rms": 0.6242942240346081, + "mirror_update_rms": 0.06731443842928729, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 4224, + "train_examples": 45000, + "train_loss": 13.869815299987794 + }, + { + "epoch": 13, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.148832145293135, + "mirror_response_residual_rms": 0.7445651127936855, + "mirror_update_rms": 0.08094001692560394, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 4576, + "train_examples": 45000, + "train_loss": 19.6516478451199 + }, + { + "epoch": 14, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14691836590168458, + "mirror_response_residual_rms": 0.8481962493905075, + "mirror_update_rms": 0.09503342386989845, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 4928, + "train_examples": 45000, + "train_loss": 25.50753713446723 + }, + { + "epoch": 15, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.14379049690829254, + "mirror_response_residual_rms": 0.9495757975526531, + "mirror_update_rms": 0.10675460737862502, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 5280, + "train_examples": 45000, + "train_loss": 30.04840303717719 + }, + { + "epoch": 16, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1507171526261422, + "mirror_response_residual_rms": 1.1510456359288477, + "mirror_update_rms": 0.12834573186914466, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 5632, + "train_examples": 45000, + "train_loss": 45.06078114624023 + }, + { + "epoch": 17, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.15367017306020433, + "mirror_response_residual_rms": 1.37932800198578, + "mirror_update_rms": 0.157787521273463, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 5984, + "train_examples": 45000, + "train_loss": 63.5289334499783 + }, + { + "epoch": 18, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.15778397220927287, + "mirror_response_residual_rms": 1.6670247928094992, + "mirror_update_rms": 0.19010092950359214, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 6336, + "train_examples": 45000, + "train_loss": 88.0200079155816 + }, + { + "epoch": 19, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16451180865244322, + "mirror_response_residual_rms": 2.07706585442247, + "mirror_update_rms": 0.23401384275961049, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 6688, + "train_examples": 45000, + "train_loss": 137.8758198594835 + }, + { + "epoch": 20, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16324027357678314, + "mirror_response_residual_rms": 2.4645421680686357, + "mirror_update_rms": 0.28651428026731585, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 7040, + "train_examples": 45000, + "train_loss": 195.62825753580728 + }, + { + "epoch": 21, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16087264351439118, + "mirror_response_residual_rms": 2.8832505643891166, + "mirror_update_rms": 0.33799926118866, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 7392, + "train_examples": 45000, + "train_loss": 242.12724807400173 + }, + { + "epoch": 22, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16066775997432384, + "mirror_response_residual_rms": 3.4065093210563533, + "mirror_update_rms": 0.40086883494305686, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 7744, + "train_examples": 45000, + "train_loss": 374.081559109158 + }, + { + "epoch": 23, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1605061240086014, + "mirror_response_residual_rms": 4.04260058210834, + "mirror_update_rms": 0.4765740674998487, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 8096, + "train_examples": 45000, + "train_loss": 482.38347734375 + }, + { + "epoch": 24, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16148118244370566, + "mirror_response_residual_rms": 4.864803373614072, + "mirror_update_rms": 0.5765041467876094, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 8448, + "train_examples": 45000, + "train_loss": 669.3851554253472 + }, + { + "epoch": 25, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16112621762216398, + "mirror_response_residual_rms": 5.7837605311443605, + "mirror_update_rms": 0.6935191516283535, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 8800, + "train_examples": 45000, + "train_loss": 945.77810703125 + }, + { + "epoch": 26, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16955177432825977, + "mirror_response_residual_rms": 7.38037095066013, + "mirror_update_rms": 0.887455653630104, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 9152, + "train_examples": 45000, + "train_loss": 1498.9955378689235 + }, + { + "epoch": 27, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.165461056050357, + "mirror_response_residual_rms": 8.671575736742273, + "mirror_update_rms": 1.0559527828793684, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 9504, + "train_examples": 45000, + "train_loss": 2073.428576931424 + }, + { + "epoch": 28, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.16891245524497428, + "mirror_response_residual_rms": 10.776266650324944, + "mirror_update_rms": 1.3449962853721882, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 9856, + "train_examples": 45000, + "train_loss": 3305.028286545139 + }, + { + "epoch": 29, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1822719349967337, + "mirror_response_residual_rms": 14.59802131644838, + "mirror_update_rms": 1.7913343699511124, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 10208, + "train_examples": 45000, + "train_loss": 5879.760441753472 + }, + { + "epoch": 30, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17733698926528618, + "mirror_response_residual_rms": 17.678691942838245, + "mirror_update_rms": 2.2343160417827197, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 10560, + "train_examples": 45000, + "train_loss": 8133.397268142361 + }, + { + "epoch": 31, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18213210933401422, + "mirror_response_residual_rms": 22.904314957885802, + "mirror_update_rms": 2.8373341960709793, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 10912, + "train_examples": 45000, + "train_loss": 13725.4761734375 + }, + { + "epoch": 32, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1789663801607258, + "mirror_response_residual_rms": 27.909721052714108, + "mirror_update_rms": 3.500229807028831, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 11264, + "train_examples": 45000, + "train_loss": 21255.801840277778 + }, + { + "epoch": 33, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1826883230505971, + "mirror_response_residual_rms": 36.14833313188745, + "mirror_update_rms": 4.585687181367318, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 11616, + "train_examples": 45000, + "train_loss": 34844.32973229167 + }, + { + "epoch": 34, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18316842271969627, + "mirror_response_residual_rms": 45.552103503047434, + "mirror_update_rms": 5.635638935034679, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 11968, + "train_examples": 45000, + "train_loss": 54909.510695833334 + }, + { + "epoch": 35, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17648959479192375, + "mirror_response_residual_rms": 54.71937135242436, + "mirror_update_rms": 7.081058243249486, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 12320, + "train_examples": 45000, + "train_loss": 78984.31330972222 + }, + { + "epoch": 36, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18068314391799956, + "mirror_response_residual_rms": 70.05446728839674, + "mirror_update_rms": 8.812665459758465, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 12672, + "train_examples": 45000, + "train_loss": 127665.53662222222 + }, + { + "epoch": 37, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17869210854684028, + "mirror_response_residual_rms": 86.35653559685802, + "mirror_update_rms": 10.983280710794133, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 13024, + "train_examples": 45000, + "train_loss": 201991.81146111112 + }, + { + "epoch": 38, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18076611558431824, + "mirror_response_residual_rms": 109.84225883439413, + "mirror_update_rms": 13.950523977016003, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 13376, + "train_examples": 45000, + "train_loss": 309671.7070666667 + }, + { + "epoch": 39, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17605196837879125, + "mirror_response_residual_rms": 133.35982368514505, + "mirror_update_rms": 17.57761647720774, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 13728, + "train_examples": 45000, + "train_loss": 465965.03088888887 + }, + { + "epoch": 40, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18244029527013597, + "mirror_response_residual_rms": 173.67093988393012, + "mirror_update_rms": 22.281607999226715, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 14080, + "train_examples": 45000, + "train_loss": 811913.9667111111 + }, + { + "epoch": 41, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18349701300469448, + "mirror_response_residual_rms": 222.03079537354787, + "mirror_update_rms": 29.00421429258447, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 14432, + "train_examples": 45000, + "train_loss": 1318171.975288889 + }, + { + "epoch": 42, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1865372831351474, + "mirror_response_residual_rms": 287.27786676741204, + "mirror_update_rms": 36.48357445045527, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 14784, + "train_examples": 45000, + "train_loss": 2149740.2739333333 + }, + { + "epoch": 43, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18663987868197604, + "mirror_response_residual_rms": 365.9685794801296, + "mirror_update_rms": 46.608857436484044, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 15136, + "train_examples": 45000, + "train_loss": 3348368.5892 + }, + { + "epoch": 44, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1879507796520401, + "mirror_response_residual_rms": 467.21583939832925, + "mirror_update_rms": 58.335997671156115, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 15488, + "train_examples": 45000, + "train_loss": 5636463.313244444 + }, + { + "epoch": 45, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1744555416960315, + "mirror_response_residual_rms": 534.8003540964359, + "mirror_update_rms": 71.101584860161, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 15840, + "train_examples": 45000, + "train_loss": 6791371.261511111 + }, + { + "epoch": 46, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18611447036962508, + "mirror_response_residual_rms": 724.0998592595789, + "mirror_update_rms": 90.09689791886062, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 16192, + "train_examples": 45000, + "train_loss": 13092151.4672 + }, + { + "epoch": 47, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1746190608773175, + "mirror_response_residual_rms": 841.7455450074359, + "mirror_update_rms": 110.74197641879132, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 16544, + "train_examples": 45000, + "train_loss": 17874979.08088889 + }, + { + "epoch": 48, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18332063396202763, + "mirror_response_residual_rms": 1106.44388714552, + "mirror_update_rms": 142.67215538580754, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 16896, + "train_examples": 45000, + "train_loss": 29886619.14311111 + }, + { + "epoch": 49, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17559674342885276, + "mirror_response_residual_rms": 1309.3478547752472, + "mirror_update_rms": 169.48276893487449, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 17248, + "train_examples": 45000, + "train_loss": 44705749.668977775 + }, + { + "epoch": 50, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.178813686974769, + "mirror_response_residual_rms": 1666.470702060352, + "mirror_update_rms": 214.71438013214353, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 17600, + "train_examples": 45000, + "train_loss": 68346633.8752 + }, + { + "epoch": 51, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18662713391866761, + "mirror_response_residual_rms": 2217.8449316645742, + "mirror_update_rms": 284.10560923075167, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 17952, + "train_examples": 45000, + "train_loss": 124116914.99235556 + }, + { + "epoch": 52, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17934205222275518, + "mirror_response_residual_rms": 2659.718713499525, + "mirror_update_rms": 343.0783791711567, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 18304, + "train_examples": 45000, + "train_loss": 172458058.5841778 + }, + { + "epoch": 53, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1859129208889452, + "mirror_response_residual_rms": 3513.790284573236, + "mirror_update_rms": 450.09763237392434, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 18656, + "train_examples": 45000, + "train_loss": 309339393.3368889 + }, + { + "epoch": 54, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18399153254195033, + "mirror_response_residual_rms": 4393.667547175527, + "mirror_update_rms": 570.2181829209991, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 19008, + "train_examples": 45000, + "train_loss": 459414734.4497778 + }, + { + "epoch": 55, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1853021813264698, + "mirror_response_residual_rms": 5590.08499909042, + "mirror_update_rms": 718.4029199901084, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 19360, + "train_examples": 45000, + "train_loss": 793463121.351111 + }, + { + "epoch": 56, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.19489280827420655, + "mirror_response_residual_rms": 7560.064761557443, + "mirror_update_rms": 968.9969079091342, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 19712, + "train_examples": 45000, + "train_loss": 1333007938.8899555 + }, + { + "epoch": 57, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17073401016694076, + "mirror_response_residual_rms": 8083.455972286784, + "mirror_update_rms": 1081.7724898096308, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 20064, + "train_examples": 45000, + "train_loss": 1532922301.4741333 + }, + { + "epoch": 58, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17630603683121485, + "mirror_response_residual_rms": 10384.860035012036, + "mirror_update_rms": 1322.3898705386234, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 20416, + "train_examples": 45000, + "train_loss": 2652798095.223467 + }, + { + "epoch": 59, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18140232516945584, + "mirror_response_residual_rms": 13574.046510492546, + "mirror_update_rms": 1755.2198073305533, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 20768, + "train_examples": 45000, + "train_loss": 4811285971.672177 + }, + { + "epoch": 60, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18453846032492838, + "mirror_response_residual_rms": 17470.456784532664, + "mirror_update_rms": 2280.3137307007123, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 21120, + "train_examples": 45000, + "train_loss": 7552221584.497778 + }, + { + "epoch": 61, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18107685799981055, + "mirror_response_residual_rms": 21497.63299855653, + "mirror_update_rms": 2873.2881216182727, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 21472, + "train_examples": 45000, + "train_loss": 10946728254.21369 + }, + { + "epoch": 62, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18102814453764157, + "mirror_response_residual_rms": 26976.778231466178, + "mirror_update_rms": 3507.690709372985, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 21824, + "train_examples": 45000, + "train_loss": 17564997953.854576 + }, + { + "epoch": 63, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1783837304563061, + "mirror_response_residual_rms": 33000.382614263886, + "mirror_update_rms": 4304.0797659547425, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 22176, + "train_examples": 45000, + "train_loss": 26833638107.81867 + }, + { + "epoch": 64, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17535460293748267, + "mirror_response_residual_rms": 40099.594777386184, + "mirror_update_rms": 5198.018456255257, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 22528, + "train_examples": 45000, + "train_loss": 41140095756.24249 + }, + { + "epoch": 65, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17961166106353832, + "mirror_response_residual_rms": 51367.058565221116, + "mirror_update_rms": 6777.709826647028, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 22880, + "train_examples": 45000, + "train_loss": 60773717620.235374 + }, + { + "epoch": 66, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1811269145069264, + "mirror_response_residual_rms": 64865.691350121044, + "mirror_update_rms": 8338.061636316976, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 23232, + "train_examples": 45000, + "train_loss": 98383998289.32835 + }, + { + "epoch": 67, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18463290853391082, + "mirror_response_residual_rms": 83802.55149915033, + "mirror_update_rms": 10957.47010951481, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 23584, + "train_examples": 45000, + "train_loss": 176490147237.79697 + }, + { + "epoch": 68, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1794876925592518, + "mirror_response_residual_rms": 101696.82013722709, + "mirror_update_rms": 13411.347290345682, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 23936, + "train_examples": 45000, + "train_loss": 255842117496.19485 + }, + { + "epoch": 69, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17822795020917803, + "mirror_response_residual_rms": 125772.15171833048, + "mirror_update_rms": 16511.972454146486, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 24288, + "train_examples": 45000, + "train_loss": 358484388522.6667 + }, + { + "epoch": 70, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.19554289081997547, + "mirror_response_residual_rms": 177640.9328548121, + "mirror_update_rms": 22584.679571557284, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 24640, + "train_examples": 45000, + "train_loss": 792783113077.1001 + }, + { + "epoch": 71, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1861383153455823, + "mirror_response_residual_rms": 213763.8259464994, + "mirror_update_rms": 28572.979543306457, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 24992, + "train_examples": 45000, + "train_loss": 1043875938761.2957 + }, + { + "epoch": 72, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1802558879375276, + "mirror_response_residual_rms": 258722.47525963813, + "mirror_update_rms": 34050.38042401327, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 25344, + "train_examples": 45000, + "train_loss": 1588843875908.5398 + }, + { + "epoch": 73, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18096636689451612, + "mirror_response_residual_rms": 324379.86535966373, + "mirror_update_rms": 42201.77580655228, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 25696, + "train_examples": 45000, + "train_loss": 2706677120683.213 + }, + { + "epoch": 74, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18419010067841654, + "mirror_response_residual_rms": 418610.1161341559, + "mirror_update_rms": 55279.74510038989, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 26048, + "train_examples": 45000, + "train_loss": 3922646842660.9097 + }, + { + "epoch": 75, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.17973940148357895, + "mirror_response_residual_rms": 508044.559060313, + "mirror_update_rms": 65859.1019107165, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 26400, + "train_examples": 45000, + "train_loss": 6108011498185.25 + }, + { + "epoch": 76, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18388618506857934, + "mirror_response_residual_rms": 657493.4999174849, + "mirror_update_rms": 85321.28164034746, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 26752, + "train_examples": 45000, + "train_loss": 10141217093344.006 + }, + { + "epoch": 77, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18735740182710184, + "mirror_response_residual_rms": 854163.0224306438, + "mirror_update_rms": 110091.04673989624, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 27104, + "train_examples": 45000, + "train_loss": 17790416377715.098 + }, + { + "epoch": 78, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.19003628168513673, + "mirror_response_residual_rms": 1119451.5912114652, + "mirror_update_rms": 147374.52377531424, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 27456, + "train_examples": 45000, + "train_loss": 30380973543280.55 + }, + { + "epoch": 79, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18090824787881385, + "mirror_response_residual_rms": 1335647.463668788, + "mirror_update_rms": 177988.60969112904, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 27808, + "train_examples": 45000, + "train_loss": 41709467618235.01 + }, + { + "epoch": 80, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1898431409649226, + "mirror_response_residual_rms": 1800206.3382607142, + "mirror_update_rms": 231471.48319929, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 28160, + "train_examples": 45000, + "train_loss": 79553879526628.11 + }, + { + "epoch": 81, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18858933860173666, + "mirror_response_residual_rms": 2290311.9855850786, + "mirror_update_rms": 304715.60732280766, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 28512, + "train_examples": 45000, + "train_loss": 134003902370149.53 + }, + { + "epoch": 82, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18903455431275143, + "mirror_response_residual_rms": 2935674.0224626358, + "mirror_update_rms": 380333.01479478984, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 28864, + "train_examples": 45000, + "train_loss": 195446795575323.66 + }, + { + "epoch": 83, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18552133326240253, + "mirror_response_residual_rms": 3649113.9615057916, + "mirror_update_rms": 472973.0840752212, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 29216, + "train_examples": 45000, + "train_loss": 328307573206507.7 + }, + { + "epoch": 84, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1871710235829224, + "mirror_response_residual_rms": 4712112.5393311195, + "mirror_update_rms": 625773.0681054839, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 29568, + "train_examples": 45000, + "train_loss": 540371527653689.5 + }, + { + "epoch": 85, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18686970229182326, + "mirror_response_residual_rms": 5959242.205373104, + "mirror_update_rms": 782316.5334189691, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 29920, + "train_examples": 45000, + "train_loss": 809226572376921.4 + }, + { + "epoch": 86, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18170413384572284, + "mirror_response_residual_rms": 7280688.028885112, + "mirror_update_rms": 943222.9128179882, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 30272, + "train_examples": 45000, + "train_loss": 1202605108566151.8 + }, + { + "epoch": 87, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18938218997417897, + "mirror_response_residual_rms": 9736101.237685226, + "mirror_update_rms": 1256339.8531903229, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 30624, + "train_examples": 45000, + "train_loss": 2290791576165842.0 + }, + { + "epoch": 88, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.18570973089994655, + "mirror_response_residual_rms": 12105269.685644234, + "mirror_update_rms": 1611442.535408771, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 30976, + "train_examples": 45000, + "train_loss": 3386755509415738.5 + }, + { + "epoch": 89, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1843602737209167, + "mirror_response_residual_rms": 15199152.999261754, + "mirror_update_rms": 1971571.6992519656, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 31328, + "train_examples": 45000, + "train_loss": 5208089187694965.0 + }, + { + "epoch": 90, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.21858175499831747, + "mirror_response_residual_rms": 23037592.17211868, + "mirror_update_rms": 2566309.0258844886, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 31680, + "train_examples": 45000, + "train_loss": 9531445722550406.0 + }, + { + "epoch": 91, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.20686517289704395, + "mirror_response_residual_rms": 26926553.356225573, + "mirror_update_rms": 3094218.4297929276, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 32032, + "train_examples": 45000, + "train_loss": 1.1028518796957628e+16 + }, + { + "epoch": 92, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.2184259488858253, + "mirror_response_residual_rms": 33706387.541730076, + "mirror_update_rms": 3494422.0886306763, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 32384, + "train_examples": 45000, + "train_loss": 1.5808196181353206e+16 + }, + { + "epoch": 93, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1900621536360849, + "mirror_response_residual_rms": 32980082.75589377, + "mirror_update_rms": 3963598.358970932, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 32736, + "train_examples": 45000, + "train_loss": 1.4888829304130642e+16 + }, + { + "epoch": 94, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.13379056089318841, + "mirror_response_residual_rms": 23828338.58614168, + "mirror_update_rms": 3886010.939256947, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 33088, + "train_examples": 45000, + "train_loss": 1.3880673278480904e+16 + }, + { + "epoch": 95, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.128798268991764, + "mirror_response_residual_rms": 23047462.405486204, + "mirror_update_rms": 3772978.128276217, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 33440, + "train_examples": 45000, + "train_loss": 1.3919191122053262e+16 + }, + { + "epoch": 96, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.1316850882560173, + "mirror_response_residual_rms": 23885031.5393649, + "mirror_update_rms": 3797964.952497794, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 33792, + "train_examples": 45000, + "train_loss": 1.2508984939937554e+16 + }, + { + "epoch": 97, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.11539434120659725, + "mirror_response_residual_rms": 20917237.293214742, + "mirror_update_rms": 3629201.827944914, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 34144, + "train_examples": 45000, + "train_loss": 1.2320241830166044e+16 + }, + { + "epoch": 98, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.10226171819833896, + "mirror_response_residual_rms": 18389204.321048595, + "mirror_update_rms": 3453547.717277361, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 34496, + "train_examples": 45000, + "train_loss": 1.0208179499010662e+16 + }, + { + "epoch": 99, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.10653423496776888, + "mirror_response_residual_rms": 19073907.00763785, + "mirror_update_rms": 3576712.541446859, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 34848, + "train_examples": 45000, + "train_loss": 1.1274575128102436e+16 + }, + { + "epoch": 100, + "lr": 0.1, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.09601334576024682, + "mirror_response_residual_rms": 16954939.118557964, + "mirror_update_rms": 3631376.9407005864, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.1, + "step": 35200, + "train_examples": 45000, + "train_loss": 9654927919423592.0 + }, + { + "epoch": 101, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.05007619556399978, + "mirror_response_residual_rms": 8678039.387749843, + "mirror_update_rms": 2200752.763350605, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 35552, + "train_examples": 45000, + "train_loss": 2685812866323736.0 + }, + { + "epoch": 102, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.025464442061644322, + "mirror_response_residual_rms": 4338527.072616534, + "mirror_update_rms": 831726.585555313, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 35904, + "train_examples": 45000, + "train_loss": 1542737949098170.5 + }, + { + "epoch": 103, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.017440496623211545, + "mirror_response_residual_rms": 2953668.1739617838, + "mirror_update_rms": 413629.8954984534, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 36256, + "train_examples": 45000, + "train_loss": 1183100001330467.0 + }, + { + "epoch": 104, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.013086733649498918, + "mirror_response_residual_rms": 2194251.548283121, + "mirror_update_rms": 257828.15337585015, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 36608, + "train_examples": 45000, + "train_loss": 968386359265154.4 + }, + { + "epoch": 105, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.01088335763231425, + "mirror_response_residual_rms": 1813165.2837789103, + "mirror_update_rms": 195048.0940523917, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 36960, + "train_examples": 45000, + "train_loss": 869776046322561.9 + }, + { + "epoch": 106, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.009499648803500215, + "mirror_response_residual_rms": 1573611.7734772088, + "mirror_update_rms": 162998.96583068467, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 37312, + "train_examples": 45000, + "train_loss": 768828992904786.4 + }, + { + "epoch": 107, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.008443350908718023, + "mirror_response_residual_rms": 1390910.5224194033, + "mirror_update_rms": 138976.35164619682, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 37664, + "train_examples": 45000, + "train_loss": 731888668104644.6 + }, + { + "epoch": 108, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.008011575043358851, + "mirror_response_residual_rms": 1312958.2721799032, + "mirror_update_rms": 128914.45254236949, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 38016, + "train_examples": 45000, + "train_loss": 713305250399594.5 + }, + { + "epoch": 109, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0073659721440511835, + "mirror_response_residual_rms": 1201779.4236550073, + "mirror_update_rms": 120465.23578049015, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 38368, + "train_examples": 45000, + "train_loss": 677757413458252.0 + }, + { + "epoch": 110, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007133456253102175, + "mirror_response_residual_rms": 1157068.9408236998, + "mirror_update_rms": 112291.41033770666, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 38720, + "train_examples": 45000, + "train_loss": 625836772680904.8 + }, + { + "epoch": 111, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007280583051060433, + "mirror_response_residual_rms": 1177446.0560974798, + "mirror_update_rms": 110508.36704757994, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 39072, + "train_examples": 45000, + "train_loss": 672107429864931.4 + }, + { + "epoch": 112, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006962275900756396, + "mirror_response_residual_rms": 1119853.1181717757, + "mirror_update_rms": 104260.1949019794, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 39424, + "train_examples": 45000, + "train_loss": 584026185917035.0 + }, + { + "epoch": 113, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007171235033341957, + "mirror_response_residual_rms": 1150059.4935755366, + "mirror_update_rms": 103543.28291815725, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 39776, + "train_examples": 45000, + "train_loss": 655910883588786.6 + }, + { + "epoch": 114, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006791146840872226, + "mirror_response_residual_rms": 1085348.125182368, + "mirror_update_rms": 100022.97957162489, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 40128, + "train_examples": 45000, + "train_loss": 594731629461728.2 + }, + { + "epoch": 115, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0066752400037565504, + "mirror_response_residual_rms": 1061156.3200544603, + "mirror_update_rms": 97680.03463975567, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 40480, + "train_examples": 45000, + "train_loss": 585738924981922.5 + }, + { + "epoch": 116, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.00687073840358863, + "mirror_response_residual_rms": 1089158.2530041405, + "mirror_update_rms": 97449.83883362441, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 40832, + "train_examples": 45000, + "train_loss": 648997796046001.5 + }, + { + "epoch": 117, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006623191988131194, + "mirror_response_residual_rms": 1044907.0573297386, + "mirror_update_rms": 94962.5233855449, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 41184, + "train_examples": 45000, + "train_loss": 627047655602394.9 + }, + { + "epoch": 118, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006828525904102637, + "mirror_response_residual_rms": 1074762.5561427027, + "mirror_update_rms": 98280.48257698271, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 41536, + "train_examples": 45000, + "train_loss": 628574778039631.1 + }, + { + "epoch": 119, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006799931982003371, + "mirror_response_residual_rms": 1064404.0593299875, + "mirror_update_rms": 95447.1747766641, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 41888, + "train_examples": 45000, + "train_loss": 589941104878201.5 + }, + { + "epoch": 120, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0067998542391554395, + "mirror_response_residual_rms": 1061315.1924613614, + "mirror_update_rms": 94584.0734345371, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 42240, + "train_examples": 45000, + "train_loss": 573994089067536.6 + }, + { + "epoch": 121, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006642465879742914, + "mirror_response_residual_rms": 1032394.358680163, + "mirror_update_rms": 92066.28875345645, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 42592, + "train_examples": 45000, + "train_loss": 568712569978850.9 + }, + { + "epoch": 122, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006595007918939228, + "mirror_response_residual_rms": 1017818.1026309867, + "mirror_update_rms": 90729.02627546371, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 42944, + "train_examples": 45000, + "train_loss": 552122170345963.9 + }, + { + "epoch": 123, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006832428471027269, + "mirror_response_residual_rms": 1053073.9852018063, + "mirror_update_rms": 91736.27698295841, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 43296, + "train_examples": 45000, + "train_loss": 618764466178810.8 + }, + { + "epoch": 124, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006809612547507064, + "mirror_response_residual_rms": 1046847.8183816015, + "mirror_update_rms": 92294.90685002417, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 43648, + "train_examples": 45000, + "train_loss": 611452914381351.2 + }, + { + "epoch": 125, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006667132503391487, + "mirror_response_residual_rms": 1021494.4585219845, + "mirror_update_rms": 92605.79022325971, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 44000, + "train_examples": 45000, + "train_loss": 581556050629586.0 + }, + { + "epoch": 126, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006782468377429878, + "mirror_response_residual_rms": 1033167.6328883163, + "mirror_update_rms": 92334.49869874583, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 44352, + "train_examples": 45000, + "train_loss": 600776144712397.2 + }, + { + "epoch": 127, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006833546552036531, + "mirror_response_residual_rms": 1038288.8922962393, + "mirror_update_rms": 92362.97052286884, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 44704, + "train_examples": 45000, + "train_loss": 631050040659038.1 + }, + { + "epoch": 128, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006990183509284284, + "mirror_response_residual_rms": 1060676.5728971122, + "mirror_update_rms": 95248.81431576378, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 45056, + "train_examples": 45000, + "train_loss": 624655222047675.4 + }, + { + "epoch": 129, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0068044943283255745, + "mirror_response_residual_rms": 1025096.6113397344, + "mirror_update_rms": 90868.9044800425, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 45408, + "train_examples": 45000, + "train_loss": 559598045456615.2 + }, + { + "epoch": 130, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.00664731148503295, + "mirror_response_residual_rms": 998640.0638799268, + "mirror_update_rms": 88580.75029270866, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 45760, + "train_examples": 45000, + "train_loss": 553175264007314.2 + }, + { + "epoch": 131, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.00727493634617216, + "mirror_response_residual_rms": 1089527.24497701, + "mirror_update_rms": 92118.74338608036, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 46112, + "train_examples": 45000, + "train_loss": 637205817216467.0 + }, + { + "epoch": 132, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007176356642018252, + "mirror_response_residual_rms": 1071260.3932451278, + "mirror_update_rms": 93108.29632830959, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 46464, + "train_examples": 45000, + "train_loss": 585200561449482.5 + }, + { + "epoch": 133, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006926766321691114, + "mirror_response_residual_rms": 1032639.9513879622, + "mirror_update_rms": 90711.73210164852, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 46816, + "train_examples": 45000, + "train_loss": 608497134244911.1 + }, + { + "epoch": 134, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007031989825841695, + "mirror_response_residual_rms": 1041955.9011630568, + "mirror_update_rms": 90427.13109136443, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 47168, + "train_examples": 45000, + "train_loss": 620365679009039.1 + }, + { + "epoch": 135, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006995993190008252, + "mirror_response_residual_rms": 1033365.2555759613, + "mirror_update_rms": 90241.4221074971, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 47520, + "train_examples": 45000, + "train_loss": 594346064258473.1 + }, + { + "epoch": 136, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007197294502746857, + "mirror_response_residual_rms": 1058573.849334281, + "mirror_update_rms": 90254.36168400788, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 47872, + "train_examples": 45000, + "train_loss": 612038576587121.9 + }, + { + "epoch": 137, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007682614039846408, + "mirror_response_residual_rms": 1126129.1386334912, + "mirror_update_rms": 95748.24352914802, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 48224, + "train_examples": 45000, + "train_loss": 626325654461824.6 + }, + { + "epoch": 138, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007144809275862734, + "mirror_response_residual_rms": 1045163.2029975861, + "mirror_update_rms": 94728.94677210202, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 48576, + "train_examples": 45000, + "train_loss": 590664627087404.8 + }, + { + "epoch": 139, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006923401842169676, + "mirror_response_residual_rms": 1009537.7472674883, + "mirror_update_rms": 91395.75034929103, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 48928, + "train_examples": 45000, + "train_loss": 568850763128580.0 + }, + { + "epoch": 140, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007964083169015453, + "mirror_response_residual_rms": 1155621.3027283526, + "mirror_update_rms": 95774.40004307772, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 49280, + "train_examples": 45000, + "train_loss": 644006964805902.9 + }, + { + "epoch": 141, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0072351711936741595, + "mirror_response_residual_rms": 1046475.7189914059, + "mirror_update_rms": 92505.73905233058, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 49632, + "train_examples": 45000, + "train_loss": 590609789821858.6 + }, + { + "epoch": 142, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0071366416802181995, + "mirror_response_residual_rms": 1028323.5082519726, + "mirror_update_rms": 90241.17121400796, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 49984, + "train_examples": 45000, + "train_loss": 562891237973289.25 + }, + { + "epoch": 143, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007187517113215339, + "mirror_response_residual_rms": 1032971.3788936894, + "mirror_update_rms": 88456.59733694022, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 50336, + "train_examples": 45000, + "train_loss": 578743017739864.6 + }, + { + "epoch": 144, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007152436765486301, + "mirror_response_residual_rms": 1022168.385118116, + "mirror_update_rms": 88797.28812404908, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 50688, + "train_examples": 45000, + "train_loss": 566067590032616.5 + }, + { + "epoch": 145, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.006981905942004935, + "mirror_response_residual_rms": 995277.1459890343, + "mirror_update_rms": 88761.07033959444, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 51040, + "train_examples": 45000, + "train_loss": 574351047832004.2 + }, + { + "epoch": 146, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.00733955937548507, + "mirror_response_residual_rms": 1043803.7987524386, + "mirror_update_rms": 89058.0615793717, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 51392, + "train_examples": 45000, + "train_loss": 576169318693417.8 + }, + { + "epoch": 147, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007256925117479931, + "mirror_response_residual_rms": 1027582.2779337709, + "mirror_update_rms": 88435.61686661586, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 51744, + "train_examples": 45000, + "train_loss": 547014509576824.06 + }, + { + "epoch": 148, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007177127243177316, + "mirror_response_residual_rms": 1012994.2162651333, + "mirror_update_rms": 87382.46114962958, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 52096, + "train_examples": 45000, + "train_loss": 532413816954229.0 + }, + { + "epoch": 149, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007089626319017847, + "mirror_response_residual_rms": 997003.6974855083, + "mirror_update_rms": 86343.38372993768, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 52448, + "train_examples": 45000, + "train_loss": 546490833460532.94 + }, + { + "epoch": 150, + "lr": 0.010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.007007133536513155, + "mirror_response_residual_rms": 980852.6251162622, + "mirror_update_rms": 86053.87791524969, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.010000000000000002, + "step": 52800, + "train_examples": 45000, + "train_loss": 536405733433724.1 + }, + { + "epoch": 151, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0032087842642215675, + "mirror_response_residual_rms": 447777.9449965618, + "mirror_update_rms": 52933.57200670927, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 53152, + "train_examples": 45000, + "train_loss": 290791142991977.94 + }, + { + "epoch": 152, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.001999295464187333, + "mirror_response_residual_rms": 279073.6974907134, + "mirror_update_rms": 23165.475315146145, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 53504, + "train_examples": 45000, + "train_loss": 255500716981762.28 + }, + { + "epoch": 153, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0017815018538181486, + "mirror_response_residual_rms": 248959.96729436552, + "mirror_update_rms": 16594.365705704, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 53856, + "train_examples": 45000, + "train_loss": 237142084347768.2 + }, + { + "epoch": 154, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0016639341173205543, + "mirror_response_residual_rms": 232207.99658755856, + "mirror_update_rms": 14970.390528111138, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 54208, + "train_examples": 45000, + "train_loss": 224787508784115.8 + }, + { + "epoch": 155, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0016099386838185669, + "mirror_response_residual_rms": 224474.44327205757, + "mirror_update_rms": 14151.976886641462, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 54560, + "train_examples": 45000, + "train_loss": 212567207284117.78 + }, + { + "epoch": 156, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0014943898185704003, + "mirror_response_residual_rms": 208407.78769944093, + "mirror_update_rms": 13113.396606462915, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 54912, + "train_examples": 45000, + "train_loss": 208400407312174.66 + }, + { + "epoch": 157, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0014775262134272362, + "mirror_response_residual_rms": 205385.44525231214, + "mirror_update_rms": 12672.589007105735, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 55264, + "train_examples": 45000, + "train_loss": 194515582748086.2 + }, + { + "epoch": 158, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0014208568165277507, + "mirror_response_residual_rms": 197479.6332734282, + "mirror_update_rms": 12243.487475586227, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 55616, + "train_examples": 45000, + "train_loss": 189595967321124.97 + }, + { + "epoch": 159, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0013475312519654994, + "mirror_response_residual_rms": 187505.40203064497, + "mirror_update_rms": 11711.669432348875, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 55968, + "train_examples": 45000, + "train_loss": 183776109209647.7 + }, + { + "epoch": 160, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0013431393270176373, + "mirror_response_residual_rms": 187143.39466572073, + "mirror_update_rms": 11320.88741566046, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 56320, + "train_examples": 45000, + "train_loss": 173374222030427.47 + }, + { + "epoch": 161, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0012856528402726945, + "mirror_response_residual_rms": 178642.99113782975, + "mirror_update_rms": 11126.734709863245, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 56672, + "train_examples": 45000, + "train_loss": 173037323713039.56 + }, + { + "epoch": 162, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0012500955155314282, + "mirror_response_residual_rms": 173963.05234524232, + "mirror_update_rms": 10750.137351449426, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 57024, + "train_examples": 45000, + "train_loss": 162943987695717.94 + }, + { + "epoch": 163, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0012274067571383946, + "mirror_response_residual_rms": 170544.54784321238, + "mirror_update_rms": 10308.406553602488, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 57376, + "train_examples": 45000, + "train_loss": 160870316729317.6 + }, + { + "epoch": 164, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0012143428054024142, + "mirror_response_residual_rms": 168541.9411038527, + "mirror_update_rms": 10026.049650677493, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 57728, + "train_examples": 45000, + "train_loss": 152615502997403.16 + }, + { + "epoch": 165, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0011618848960382778, + "mirror_response_residual_rms": 161249.47841337245, + "mirror_update_rms": 9877.76319917752, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 58080, + "train_examples": 45000, + "train_loss": 147715078304539.9 + }, + { + "epoch": 166, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0011709939764778443, + "mirror_response_residual_rms": 162409.74849408612, + "mirror_update_rms": 9581.047165424463, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 58432, + "train_examples": 45000, + "train_loss": 149022314113729.97 + }, + { + "epoch": 167, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.00114368928852084, + "mirror_response_residual_rms": 158723.66304442205, + "mirror_update_rms": 9274.600017548708, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 58784, + "train_examples": 45000, + "train_loss": 145975274601662.97 + }, + { + "epoch": 168, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0011113454636964505, + "mirror_response_residual_rms": 153535.55047773395, + "mirror_update_rms": 9169.623612582436, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 59136, + "train_examples": 45000, + "train_loss": 140674355640815.53 + }, + { + "epoch": 169, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0010955640187658362, + "mirror_response_residual_rms": 151636.68171368883, + "mirror_update_rms": 9028.461506247086, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 59488, + "train_examples": 45000, + "train_loss": 138681999363460.3 + }, + { + "epoch": 170, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0011083108631044982, + "mirror_response_residual_rms": 153273.4338426599, + "mirror_update_rms": 8847.45158370748, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 59840, + "train_examples": 45000, + "train_loss": 134086743509958.66 + }, + { + "epoch": 171, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.001059070866597855, + "mirror_response_residual_rms": 146400.78252848756, + "mirror_update_rms": 8638.841519859418, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 60192, + "train_examples": 45000, + "train_loss": 130867960700690.61 + }, + { + "epoch": 172, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0010782376483572434, + "mirror_response_residual_rms": 148982.80637432565, + "mirror_update_rms": 8500.595883124372, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 60544, + "train_examples": 45000, + "train_loss": 131899575170941.3 + }, + { + "epoch": 173, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.001025326273471645, + "mirror_response_residual_rms": 141895.38581659968, + "mirror_update_rms": 8448.631889640494, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 60896, + "train_examples": 45000, + "train_loss": 126561423696354.05 + }, + { + "epoch": 174, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0010562072572070209, + "mirror_response_residual_rms": 145845.37717105175, + "mirror_update_rms": 8471.085631430662, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 61248, + "train_examples": 45000, + "train_loss": 124120458756020.27 + }, + { + "epoch": 175, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0010165783017837598, + "mirror_response_residual_rms": 140287.6153044427, + "mirror_update_rms": 8216.515302120677, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 61600, + "train_examples": 45000, + "train_loss": 122149353463835.67 + }, + { + "epoch": 176, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009861995894133544, + "mirror_response_residual_rms": 135997.74408898997, + "mirror_update_rms": 7994.437876091781, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 61952, + "train_examples": 45000, + "train_loss": 119131704066991.72 + }, + { + "epoch": 177, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0010089913501434918, + "mirror_response_residual_rms": 139091.17556596536, + "mirror_update_rms": 8007.664973720693, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 62304, + "train_examples": 45000, + "train_loss": 118752039003476.61 + }, + { + "epoch": 178, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009991471698646422, + "mirror_response_residual_rms": 137830.24079490898, + "mirror_update_rms": 7845.982322296591, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 62656, + "train_examples": 45000, + "train_loss": 116937039180207.81 + }, + { + "epoch": 179, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009777638864413196, + "mirror_response_residual_rms": 134920.04129773405, + "mirror_update_rms": 7699.635825450148, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 63008, + "train_examples": 45000, + "train_loss": 115498679129004.08 + }, + { + "epoch": 180, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009629198876278138, + "mirror_response_residual_rms": 132860.51200841728, + "mirror_update_rms": 7640.070607115113, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 63360, + "train_examples": 45000, + "train_loss": 110589073182978.5 + }, + { + "epoch": 181, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009544134332446694, + "mirror_response_residual_rms": 131206.3883222477, + "mirror_update_rms": 7493.416145239424, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 63712, + "train_examples": 45000, + "train_loss": 107961896251995.66 + }, + { + "epoch": 182, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009379672638411429, + "mirror_response_residual_rms": 129260.96492484286, + "mirror_update_rms": 7425.527523645123, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 64064, + "train_examples": 45000, + "train_loss": 105197935579714.9 + }, + { + "epoch": 183, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.000977036531632824, + "mirror_response_residual_rms": 134436.5471177028, + "mirror_update_rms": 7443.531530709466, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 64416, + "train_examples": 45000, + "train_loss": 108604489967261.73 + }, + { + "epoch": 184, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009384160472791108, + "mirror_response_residual_rms": 129188.49014012153, + "mirror_update_rms": 7296.443772341943, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 64768, + "train_examples": 45000, + "train_loss": 109384364098019.52 + }, + { + "epoch": 185, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009414905730334572, + "mirror_response_residual_rms": 129269.97882887031, + "mirror_update_rms": 7243.7012446729705, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 65120, + "train_examples": 45000, + "train_loss": 103255089672082.05 + }, + { + "epoch": 186, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.000932192632579979, + "mirror_response_residual_rms": 128013.12664497542, + "mirror_update_rms": 7146.149532650844, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 65472, + "train_examples": 45000, + "train_loss": 102638880865526.31 + }, + { + "epoch": 187, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009103720910796426, + "mirror_response_residual_rms": 124976.02616910689, + "mirror_update_rms": 7148.405145932297, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 65824, + "train_examples": 45000, + "train_loss": 98623481536018.3 + }, + { + "epoch": 188, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009235896524995814, + "mirror_response_residual_rms": 126771.72839843947, + "mirror_update_rms": 7089.127334086657, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 66176, + "train_examples": 45000, + "train_loss": 97190608395013.69 + }, + { + "epoch": 189, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.000918707577524986, + "mirror_response_residual_rms": 126163.4560925949, + "mirror_update_rms": 7040.906495507836, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 66528, + "train_examples": 45000, + "train_loss": 100718257142543.89 + }, + { + "epoch": 190, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009150105731567869, + "mirror_response_residual_rms": 125399.70188986123, + "mirror_update_rms": 7085.229258096326, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 66880, + "train_examples": 45000, + "train_loss": 100124271651050.11 + }, + { + "epoch": 191, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009168686247955499, + "mirror_response_residual_rms": 125690.86517330584, + "mirror_update_rms": 6912.006706956782, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 67232, + "train_examples": 45000, + "train_loss": 96021660878026.25 + }, + { + "epoch": 192, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008978633099696281, + "mirror_response_residual_rms": 123156.36110672857, + "mirror_update_rms": 6793.551535886159, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 67584, + "train_examples": 45000, + "train_loss": 96242815941150.12 + }, + { + "epoch": 193, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008827945562784333, + "mirror_response_residual_rms": 121085.07525103634, + "mirror_update_rms": 6824.396519630675, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 67936, + "train_examples": 45000, + "train_loss": 90978474028346.03 + }, + { + "epoch": 194, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009324719244620326, + "mirror_response_residual_rms": 127985.53551857443, + "mirror_update_rms": 6947.703788473882, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 68288, + "train_examples": 45000, + "train_loss": 96810687187403.84 + }, + { + "epoch": 195, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0009198801757618173, + "mirror_response_residual_rms": 125902.66336070251, + "mirror_update_rms": 6847.628727216657, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 68640, + "train_examples": 45000, + "train_loss": 94776781496344.77 + }, + { + "epoch": 196, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008748863621404889, + "mirror_response_residual_rms": 119901.70220166737, + "mirror_update_rms": 6681.125528773386, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 68992, + "train_examples": 45000, + "train_loss": 94696944802481.22 + }, + { + "epoch": 197, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008890893212298624, + "mirror_response_residual_rms": 121926.22679854333, + "mirror_update_rms": 6669.037733824636, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 69344, + "train_examples": 45000, + "train_loss": 91934331761194.14 + }, + { + "epoch": 198, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008945555896561919, + "mirror_response_residual_rms": 122263.97020262144, + "mirror_update_rms": 6600.694677847766, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 69696, + "train_examples": 45000, + "train_loss": 91486208920160.39 + }, + { + "epoch": 199, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008649277129327108, + "mirror_response_residual_rms": 118035.00068733803, + "mirror_update_rms": 6560.013784059727, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 70048, + "train_examples": 45000, + "train_loss": 87027075818455.05 + }, + { + "epoch": 200, + "lr": 0.0010000000000000002, + "mirror": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.0008524223930613697, + "mirror_response_residual_rms": 116590.40100360393, + "mirror_update_rms": 6515.713952715815, + "readout_batch_size": 1024.0 + }, + "output_lr": 0.0010000000000000002, + "step": 70400, + "train_examples": 45000, + "train_loss": 89503349143584.4 + } + ], + "evaluation_protocol": { + "test_evaluations": 0, + "test_used_for_selection": false, + "validation_evaluations": 1 + }, + "final": { + "accuracy": 0.1, + "epoch": 200, + "evaluation_split": "validation", + "finite": false, + "loss": NaN, + "step": 70400 + }, + "hardware": { + "cuda_device_name": "NVIDIA GeForce GTX 1080", + "cuda_visible_devices": "5", + "device": "cuda", + "device_total_memory_bytes": 8507949056, + "peak_memory_allocated_bytes": 2161716736, + "peak_memory_reserved_bytes": 2883584000, + "torch_version": "2.3.1+cu118" + }, + "mirror_warmup": { + "first": { + "conv_batch_size": 1, + "mirror_response_residual_fraction": 1.4124661316345333, + "mirror_response_residual_rms": 1.90927576279183, + "mirror_update_rms": 0.25520769706611623, + "readout_batch_size": 1024 + }, + "last": { + "conv_batch_size": 1, + "mirror_response_residual_fraction": 0.26271250078043745, + "mirror_response_residual_rms": 0.3557588547676734, + "mirror_update_rms": 0.0907495165488641, + "readout_batch_size": 1024 + }, + "mean": { + "conv_batch_size": 1.0, + "mirror_response_residual_fraction": 0.6719071790962137, + "mirror_response_residual_rms": 0.9100978838204868, + "mirror_update_rms": 0.15685383656211757, + "readout_batch_size": 1024.0 + }, + "steps": 20 + }, + "protocol_family": "oral_a_cifar_local_resnet_development", + "provenance": { + "git_commit": "8a5c1045a43b1f91fd46b2fcc044b032cb14718c", + "git_tracked_dirty": false + }, + "schema_version": 1, + "split": { + "cifar_source_files": [ + { + "bytes": 31035704, + "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_1", + "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd" + }, + { + "bytes": 31035320, + "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_2", + "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8" + }, + { + "bytes": 31035999, + "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_3", + "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9" + }, + { + "bytes": 31035696, + "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_4", + "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b" + }, + { + "bytes": 31035623, + "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_5", + "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154" + }, + { + "bytes": 31035526, + "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/test_batch", + "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831" + } + ], + "dataset": "cifar10", + "evaluation_split": "validation", + "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" + }, + "timing": { + "apical_warmup_wall_s": 0.0, + "evaluation_wall_s": 0.3643169403076172, + "mirror_warmup_wall_s": 0.564356803894043, + "predictor_warmup_wall_s": 0.0, + "timing_excludes_data_loading_hashing_and_model_construction": true, + "total_timed_wall_s": 4057.1195838451385, + "train_wall_s": 4056.162163734436 + }, + "work": { + "apical_macs_per_example": 40108672, + "causal_scalar_observations": 0, + "components": { + "apical_projection_macs": 360978048000000, + "apical_regression_macs": 0, + "bp_reverse_macs_estimate": 0, + "local_weight_correlation_macs": 364959360000000, + "mirror_feedback_prediction_macs": 180174192640, + "mirror_local_correlation_macs": 180174192640, + "mirror_response_macs": 180174192640, + "ordinary_forward_macs": 364959360000000, + "perturbation_forward_macs": 0, + "warmup_clean_forward_macs": 0 + }, + "definition": "multiply-accumulates in conv/linear maps; one local weight correlation equals one forward-weight MAC count; BP reverse is estimated as one weight-gradient plus one activation-gradient convolution per forward convolution; elementwise nonlinearities and optimizer arithmetic excluded", + "forward_macs_per_example": 40551040, + "logical_batch_loss_queries": 0, + "mirror_probe_examples": 4420, + "mirror_readout_probe_examples": 4526080, + "per_example_cross_entropy_terms": 0, + "total_clean_forward_examples": 9000000, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1091437290577920 + } +} diff --git a/results/residual_mirror_full_gate.json b/results/residual_mirror_full_gate.json new file mode 100644 index 0000000..a515042 --- /dev/null +++ b/results/residual_mirror_full_gate.json @@ -0,0 +1,35 @@ +{ + "checks": { + "accuracy_at_least_0.88": false, + "early_alignment_at_least_0.50": true, + "finite": false, + "macs_at_most_1.15x_bp": true, + "zero_task_loss_queries": true + }, + "confirmation_test_seeds_touched": false, + "innovation_experiment_opened": false, + "metrics": { + "accuracy": 0.1, + "all_layer_alignment": 0.8487377637311032, + "bp_accuracy": 0.9162, + "bp_total_macs": 1094878080000000, + "early_third_alignment": 0.8774353464444479, + "finite": false, + "logical_batch_loss_queries": 0, + "loss": NaN, + "mac_ratio_to_bp": 0.9968573766477451, + "max_feedback_forward_norm_ratio": 1.0024465322494507, + "mean_feedback_forward_cosine": 0.999997964030818, + "min_feedback_forward_norm_ratio": 1.000157356262207, + "mirror_events": 4420, + "peak_memory_allocated_bytes": 2161716736, + "source_commit": "8a5c1045a43b1f91fd46b2fcc044b032cb14718c", + "total_macs": 1091437290577920, + "wall_s": 4057.1195838451385 + }, + "protocol": "residual_response_mirror_full_v1", + "review_score_after": 5, + "review_score_before": 5, + "score_change_rule": "inherited full baseline cannot raise score", + "status": "failed" +} |
