diff options
Diffstat (limited to 'results/oral_a_dynamic_scaling_v2/bp_d32_s10.json')
| -rw-r--r-- | results/oral_a_dynamic_scaling_v2/bp_d32_s10.json | 1986 |
1 files changed, 1986 insertions, 0 deletions
diff --git a/results/oral_a_dynamic_scaling_v2/bp_d32_s10.json b/results/oral_a_dynamic_scaling_v2/bp_d32_s10.json new file mode 100644 index 0000000..37bcaae --- /dev/null +++ b/results/oral_a_dynamic_scaling_v2/bp_d32_s10.json @@ -0,0 +1,1986 @@ +{ + "architecture": { + "adaptive_apical_parameters": 0, + "adaptive_feedback_parameters": 0, + "base_width": 16, + "blocks_per_stage": 5, + "bn_eps": 1e-05, + "bn_momentum": 0.1, + "depth": 32, + "family": "CIFAR 6n+2 ResNet, option-A shortcuts", + "fixed_feedback_parameters": 0, + "fixed_traffic_coefficients": 0, + "forward_parameters": 464154, + "hidden_shapes": [ + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 16, + 32, + 32 + ], + [ + 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 + ], + [ + 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 + ], + [ + 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": 0, + "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": 32, + "device": "cuda", + "epochs": 200, + "eta_A": 0.01, + "eta_P": 0.01, + "eval_every": 0, + "eval_split": "test", + "learn_P": 0, + "loader_seed": 10, + "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": 0, + "mode": "bp", + "momentum": 0.9, + "neutral_projection": 0, + "normalization": "batchnorm", + "nuisance_scale": 0.0, + "out": "results/oral_a_dynamic_scaling_v2/bp_d32_s10.json", + "output_lr": null, + "pert_directions": 1, + "pert_every": 4, + "pert_sigma": 0.01, + "perturb_seed": 1000, + "predictor_every": 0, + "predictor_mode": "nlms", + "predictor_warmup_steps": 0, + "residual_scale": null, + "seed": 10, + "split_seed": 2027, + "traffic_calibration_examples": 64, + "traffic_ratio": 4.0, + "traffic_rule": "innovation", + "traffic_seed": 4000, + "train_limit": 0, + "use_residual": 1, + "val_examples": 0, + "vectorizer_mode": "spatial_template", + "warmup_epochs": 0, + "weight_decay": 0.0001, + "weight_scale": 1.0, + "width": 16 + }, + "calibration_metric_space": null, + "counters": { + "apical_warmup_examples": 0, + "calibration_event_examples": 0, + "causal_scalar_observations": 0, + "logical_batch_loss_queries": 0, + "mirror_conv_examples": 0, + "mirror_events": 0, + "mirror_readout_examples": 0, + "neutral_projection_examples": 0, + "ordinary_examples": 10000000, + "per_example_loss_terms": 0, + "perturbation_events": 0, + "perturbation_forward_examples": 0, + "predictor_update_examples": 0, + "predictor_warmup_examples": 0, + "traffic_audit_examples": 0, + "traffic_calibration_examples": 0 + }, + "diagnostics": null, + "epochs": [ + { + "epoch": 1, + "lr": 0.1, + "output_lr": 0.1, + "step": 391, + "train_examples": 50000, + "train_loss": 2.0054121128463747 + }, + { + "epoch": 2, + "lr": 0.1, + "output_lr": 0.1, + "step": 782, + "train_examples": 50000, + "train_loss": 1.4379345008850097 + }, + { + "epoch": 3, + "lr": 0.1, + "output_lr": 0.1, + "step": 1173, + "train_examples": 50000, + "train_loss": 1.1155388502883912 + }, + { + "epoch": 4, + "lr": 0.1, + "output_lr": 0.1, + "step": 1564, + "train_examples": 50000, + "train_loss": 0.9032726857757568 + }, + { + "epoch": 5, + "lr": 0.1, + "output_lr": 0.1, + "step": 1955, + "train_examples": 50000, + "train_loss": 0.7667027964401245 + }, + { + "epoch": 6, + "lr": 0.1, + "output_lr": 0.1, + "step": 2346, + "train_examples": 50000, + "train_loss": 0.676452765827179 + }, + { + "epoch": 7, + "lr": 0.1, + "output_lr": 0.1, + "step": 2737, + "train_examples": 50000, + "train_loss": 0.6164279955673218 + }, + { + "epoch": 8, + "lr": 0.1, + "output_lr": 0.1, + "step": 3128, + "train_examples": 50000, + "train_loss": 0.5678001858711242 + }, + { + "epoch": 9, + "lr": 0.1, + "output_lr": 0.1, + "step": 3519, + "train_examples": 50000, + "train_loss": 0.5277435734176635 + }, + { + "epoch": 10, + "lr": 0.1, + "output_lr": 0.1, + "step": 3910, + "train_examples": 50000, + "train_loss": 0.4982360699176788 + }, + { + "epoch": 11, + "lr": 0.1, + "output_lr": 0.1, + "step": 4301, + "train_examples": 50000, + "train_loss": 0.47352811789512633 + }, + { + "epoch": 12, + "lr": 0.1, + "output_lr": 0.1, + "step": 4692, + "train_examples": 50000, + "train_loss": 0.4530127330207825 + }, + { + "epoch": 13, + "lr": 0.1, + "output_lr": 0.1, + "step": 5083, + "train_examples": 50000, + "train_loss": 0.43370039414405825 + }, + { + "epoch": 14, + "lr": 0.1, + "output_lr": 0.1, + "step": 5474, + "train_examples": 50000, + "train_loss": 0.41464673620223996 + }, + { + "epoch": 15, + "lr": 0.1, + "output_lr": 0.1, + "step": 5865, + "train_examples": 50000, + "train_loss": 0.40234276021003723 + }, + { + "epoch": 16, + "lr": 0.1, + "output_lr": 0.1, + "step": 6256, + "train_examples": 50000, + "train_loss": 0.38951607427597046 + }, + { + "epoch": 17, + "lr": 0.1, + "output_lr": 0.1, + "step": 6647, + "train_examples": 50000, + "train_loss": 0.3729274450969696 + }, + { + "epoch": 18, + "lr": 0.1, + "output_lr": 0.1, + "step": 7038, + "train_examples": 50000, + "train_loss": 0.36208275164604187 + }, + { + "epoch": 19, + "lr": 0.1, + "output_lr": 0.1, + "step": 7429, + "train_examples": 50000, + "train_loss": 0.35236940656661986 + }, + { + "epoch": 20, + "lr": 0.1, + "output_lr": 0.1, + "step": 7820, + "train_examples": 50000, + "train_loss": 0.3433354270744324 + }, + { + "epoch": 21, + "lr": 0.1, + "output_lr": 0.1, + "step": 8211, + "train_examples": 50000, + "train_loss": 0.3330806070423126 + }, + { + "epoch": 22, + "lr": 0.1, + "output_lr": 0.1, + "step": 8602, + "train_examples": 50000, + "train_loss": 0.32628141145706174 + }, + { + "epoch": 23, + "lr": 0.1, + "output_lr": 0.1, + "step": 8993, + "train_examples": 50000, + "train_loss": 0.3165073899269104 + }, + { + "epoch": 24, + "lr": 0.1, + "output_lr": 0.1, + "step": 9384, + "train_examples": 50000, + "train_loss": 0.3097959193992615 + }, + { + "epoch": 25, + "lr": 0.1, + "output_lr": 0.1, + "step": 9775, + "train_examples": 50000, + "train_loss": 0.30136838003158567 + }, + { + "epoch": 26, + "lr": 0.1, + "output_lr": 0.1, + "step": 10166, + "train_examples": 50000, + "train_loss": 0.3010846883106232 + }, + { + "epoch": 27, + "lr": 0.1, + "output_lr": 0.1, + "step": 10557, + "train_examples": 50000, + "train_loss": 0.2891464224052429 + }, + { + "epoch": 28, + "lr": 0.1, + "output_lr": 0.1, + "step": 10948, + "train_examples": 50000, + "train_loss": 0.28535775566101074 + }, + { + "epoch": 29, + "lr": 0.1, + "output_lr": 0.1, + "step": 11339, + "train_examples": 50000, + "train_loss": 0.2811079977703094 + }, + { + "epoch": 30, + "lr": 0.1, + "output_lr": 0.1, + "step": 11730, + "train_examples": 50000, + "train_loss": 0.27119437694072723 + }, + { + "epoch": 31, + "lr": 0.1, + "output_lr": 0.1, + "step": 12121, + "train_examples": 50000, + "train_loss": 0.27670178437232973 + }, + { + "epoch": 32, + "lr": 0.1, + "output_lr": 0.1, + "step": 12512, + "train_examples": 50000, + "train_loss": 0.26804082042694094 + }, + { + "epoch": 33, + "lr": 0.1, + "output_lr": 0.1, + "step": 12903, + "train_examples": 50000, + "train_loss": 0.25923018002510073 + }, + { + "epoch": 34, + "lr": 0.1, + "output_lr": 0.1, + "step": 13294, + "train_examples": 50000, + "train_loss": 0.2588696497821808 + }, + { + "epoch": 35, + "lr": 0.1, + "output_lr": 0.1, + "step": 13685, + "train_examples": 50000, + "train_loss": 0.2541264269256592 + }, + { + "epoch": 36, + "lr": 0.1, + "output_lr": 0.1, + "step": 14076, + "train_examples": 50000, + "train_loss": 0.24870991152763366 + }, + { + "epoch": 37, + "lr": 0.1, + "output_lr": 0.1, + "step": 14467, + "train_examples": 50000, + "train_loss": 0.2512641797447205 + }, + { + "epoch": 38, + "lr": 0.1, + "output_lr": 0.1, + "step": 14858, + "train_examples": 50000, + "train_loss": 0.2503001854991913 + }, + { + "epoch": 39, + "lr": 0.1, + "output_lr": 0.1, + "step": 15249, + "train_examples": 50000, + "train_loss": 0.24354838258743286 + }, + { + "epoch": 40, + "lr": 0.1, + "output_lr": 0.1, + "step": 15640, + "train_examples": 50000, + "train_loss": 0.24228336567401887 + }, + { + "epoch": 41, + "lr": 0.1, + "output_lr": 0.1, + "step": 16031, + "train_examples": 50000, + "train_loss": 0.23563844622135163 + }, + { + "epoch": 42, + "lr": 0.1, + "output_lr": 0.1, + "step": 16422, + "train_examples": 50000, + "train_loss": 0.2350678129386902 + }, + { + "epoch": 43, + "lr": 0.1, + "output_lr": 0.1, + "step": 16813, + "train_examples": 50000, + "train_loss": 0.23324379225730896 + }, + { + "epoch": 44, + "lr": 0.1, + "output_lr": 0.1, + "step": 17204, + "train_examples": 50000, + "train_loss": 0.22955391101837158 + }, + { + "epoch": 45, + "lr": 0.1, + "output_lr": 0.1, + "step": 17595, + "train_examples": 50000, + "train_loss": 0.23024302867889404 + }, + { + "epoch": 46, + "lr": 0.1, + "output_lr": 0.1, + "step": 17986, + "train_examples": 50000, + "train_loss": 0.23181788200378417 + }, + { + "epoch": 47, + "lr": 0.1, + "output_lr": 0.1, + "step": 18377, + "train_examples": 50000, + "train_loss": 0.22251955399513246 + }, + { + "epoch": 48, + "lr": 0.1, + "output_lr": 0.1, + "step": 18768, + "train_examples": 50000, + "train_loss": 0.22411165699958802 + }, + { + "epoch": 49, + "lr": 0.1, + "output_lr": 0.1, + "step": 19159, + "train_examples": 50000, + "train_loss": 0.21793724314689636 + }, + { + "epoch": 50, + "lr": 0.1, + "output_lr": 0.1, + "step": 19550, + "train_examples": 50000, + "train_loss": 0.21977635389328004 + }, + { + "epoch": 51, + "lr": 0.1, + "output_lr": 0.1, + "step": 19941, + "train_examples": 50000, + "train_loss": 0.21169228483200073 + }, + { + "epoch": 52, + "lr": 0.1, + "output_lr": 0.1, + "step": 20332, + "train_examples": 50000, + "train_loss": 0.21168184700012208 + }, + { + "epoch": 53, + "lr": 0.1, + "output_lr": 0.1, + "step": 20723, + "train_examples": 50000, + "train_loss": 0.21563953169822692 + }, + { + "epoch": 54, + "lr": 0.1, + "output_lr": 0.1, + "step": 21114, + "train_examples": 50000, + "train_loss": 0.2104552047586441 + }, + { + "epoch": 55, + "lr": 0.1, + "output_lr": 0.1, + "step": 21505, + "train_examples": 50000, + "train_loss": 0.20921339990615845 + }, + { + "epoch": 56, + "lr": 0.1, + "output_lr": 0.1, + "step": 21896, + "train_examples": 50000, + "train_loss": 0.21239697771072388 + }, + { + "epoch": 57, + "lr": 0.1, + "output_lr": 0.1, + "step": 22287, + "train_examples": 50000, + "train_loss": 0.20206186538696289 + }, + { + "epoch": 58, + "lr": 0.1, + "output_lr": 0.1, + "step": 22678, + "train_examples": 50000, + "train_loss": 0.20385388317108155 + }, + { + "epoch": 59, + "lr": 0.1, + "output_lr": 0.1, + "step": 23069, + "train_examples": 50000, + "train_loss": 0.20894759988307954 + }, + { + "epoch": 60, + "lr": 0.1, + "output_lr": 0.1, + "step": 23460, + "train_examples": 50000, + "train_loss": 0.20231014974594116 + }, + { + "epoch": 61, + "lr": 0.1, + "output_lr": 0.1, + "step": 23851, + "train_examples": 50000, + "train_loss": 0.2059004350709915 + }, + { + "epoch": 62, + "lr": 0.1, + "output_lr": 0.1, + "step": 24242, + "train_examples": 50000, + "train_loss": 0.20370832619190216 + }, + { + "epoch": 63, + "lr": 0.1, + "output_lr": 0.1, + "step": 24633, + "train_examples": 50000, + "train_loss": 0.19682449151039125 + }, + { + "epoch": 64, + "lr": 0.1, + "output_lr": 0.1, + "step": 25024, + "train_examples": 50000, + "train_loss": 0.19642020659923554 + }, + { + "epoch": 65, + "lr": 0.1, + "output_lr": 0.1, + "step": 25415, + "train_examples": 50000, + "train_loss": 0.20093582691192627 + }, + { + "epoch": 66, + "lr": 0.1, + "output_lr": 0.1, + "step": 25806, + "train_examples": 50000, + "train_loss": 0.19774235584497452 + }, + { + "epoch": 67, + "lr": 0.1, + "output_lr": 0.1, + "step": 26197, + "train_examples": 50000, + "train_loss": 0.18971705436706543 + }, + { + "epoch": 68, + "lr": 0.1, + "output_lr": 0.1, + "step": 26588, + "train_examples": 50000, + "train_loss": 0.19689360201358796 + }, + { + "epoch": 69, + "lr": 0.1, + "output_lr": 0.1, + "step": 26979, + "train_examples": 50000, + "train_loss": 0.19998873517513274 + }, + { + "epoch": 70, + "lr": 0.1, + "output_lr": 0.1, + "step": 27370, + "train_examples": 50000, + "train_loss": 0.19119584038257598 + }, + { + "epoch": 71, + "lr": 0.1, + "output_lr": 0.1, + "step": 27761, + "train_examples": 50000, + "train_loss": 0.1885338589167595 + }, + { + "epoch": 72, + "lr": 0.1, + "output_lr": 0.1, + "step": 28152, + "train_examples": 50000, + "train_loss": 0.19239829563617705 + }, + { + "epoch": 73, + "lr": 0.1, + "output_lr": 0.1, + "step": 28543, + "train_examples": 50000, + "train_loss": 0.1873906142807007 + }, + { + "epoch": 74, + "lr": 0.1, + "output_lr": 0.1, + "step": 28934, + "train_examples": 50000, + "train_loss": 0.19268658789157866 + }, + { + "epoch": 75, + "lr": 0.1, + "output_lr": 0.1, + "step": 29325, + "train_examples": 50000, + "train_loss": 0.1842356293296814 + }, + { + "epoch": 76, + "lr": 0.1, + "output_lr": 0.1, + "step": 29716, + "train_examples": 50000, + "train_loss": 0.18621056265830993 + }, + { + "epoch": 77, + "lr": 0.1, + "output_lr": 0.1, + "step": 30107, + "train_examples": 50000, + "train_loss": 0.1835704484939575 + }, + { + "epoch": 78, + "lr": 0.1, + "output_lr": 0.1, + "step": 30498, + "train_examples": 50000, + "train_loss": 0.18719533890724183 + }, + { + "epoch": 79, + "lr": 0.1, + "output_lr": 0.1, + "step": 30889, + "train_examples": 50000, + "train_loss": 0.17780928714752198 + }, + { + "epoch": 80, + "lr": 0.1, + "output_lr": 0.1, + "step": 31280, + "train_examples": 50000, + "train_loss": 0.18500971989631654 + }, + { + "epoch": 81, + "lr": 0.1, + "output_lr": 0.1, + "step": 31671, + "train_examples": 50000, + "train_loss": 0.18152961041212082 + }, + { + "epoch": 82, + "lr": 0.1, + "output_lr": 0.1, + "step": 32062, + "train_examples": 50000, + "train_loss": 0.18043139296054841 + }, + { + "epoch": 83, + "lr": 0.1, + "output_lr": 0.1, + "step": 32453, + "train_examples": 50000, + "train_loss": 0.18088321146011352 + }, + { + "epoch": 84, + "lr": 0.1, + "output_lr": 0.1, + "step": 32844, + "train_examples": 50000, + "train_loss": 0.18133888685703278 + }, + { + "epoch": 85, + "lr": 0.1, + "output_lr": 0.1, + "step": 33235, + "train_examples": 50000, + "train_loss": 0.18393014801502228 + }, + { + "epoch": 86, + "lr": 0.1, + "output_lr": 0.1, + "step": 33626, + "train_examples": 50000, + "train_loss": 0.17943879672527313 + }, + { + "epoch": 87, + "lr": 0.1, + "output_lr": 0.1, + "step": 34017, + "train_examples": 50000, + "train_loss": 0.1776360273218155 + }, + { + "epoch": 88, + "lr": 0.1, + "output_lr": 0.1, + "step": 34408, + "train_examples": 50000, + "train_loss": 0.17978584024190902 + }, + { + "epoch": 89, + "lr": 0.1, + "output_lr": 0.1, + "step": 34799, + "train_examples": 50000, + "train_loss": 0.1760036921596527 + }, + { + "epoch": 90, + "lr": 0.1, + "output_lr": 0.1, + "step": 35190, + "train_examples": 50000, + "train_loss": 0.18229618550300597 + }, + { + "epoch": 91, + "lr": 0.1, + "output_lr": 0.1, + "step": 35581, + "train_examples": 50000, + "train_loss": 0.17563352034807206 + }, + { + "epoch": 92, + "lr": 0.1, + "output_lr": 0.1, + "step": 35972, + "train_examples": 50000, + "train_loss": 0.17081233749866487 + }, + { + "epoch": 93, + "lr": 0.1, + "output_lr": 0.1, + "step": 36363, + "train_examples": 50000, + "train_loss": 0.17763610350608824 + }, + { + "epoch": 94, + "lr": 0.1, + "output_lr": 0.1, + "step": 36754, + "train_examples": 50000, + "train_loss": 0.17541570212841034 + }, + { + "epoch": 95, + "lr": 0.1, + "output_lr": 0.1, + "step": 37145, + "train_examples": 50000, + "train_loss": 0.16629342665672303 + }, + { + "epoch": 96, + "lr": 0.1, + "output_lr": 0.1, + "step": 37536, + "train_examples": 50000, + "train_loss": 0.17705314165115357 + }, + { + "epoch": 97, + "lr": 0.1, + "output_lr": 0.1, + "step": 37927, + "train_examples": 50000, + "train_loss": 0.17434777067661286 + }, + { + "epoch": 98, + "lr": 0.1, + "output_lr": 0.1, + "step": 38318, + "train_examples": 50000, + "train_loss": 0.16857723032474517 + }, + { + "epoch": 99, + "lr": 0.1, + "output_lr": 0.1, + "step": 38709, + "train_examples": 50000, + "train_loss": 0.17178690128803253 + }, + { + "epoch": 100, + "lr": 0.1, + "output_lr": 0.1, + "step": 39100, + "train_examples": 50000, + "train_loss": 0.17055641087055207 + }, + { + "epoch": 101, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 39491, + "train_examples": 50000, + "train_loss": 0.1005971980381012 + }, + { + "epoch": 102, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 39882, + "train_examples": 50000, + "train_loss": 0.07781971687793732 + }, + { + "epoch": 103, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 40273, + "train_examples": 50000, + "train_loss": 0.06829904849290848 + }, + { + "epoch": 104, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 40664, + "train_examples": 50000, + "train_loss": 0.06221961904287338 + }, + { + "epoch": 105, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 41055, + "train_examples": 50000, + "train_loss": 0.05549530576705933 + }, + { + "epoch": 106, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 41446, + "train_examples": 50000, + "train_loss": 0.05217072459459305 + }, + { + "epoch": 107, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 41837, + "train_examples": 50000, + "train_loss": 0.04782960817337036 + }, + { + "epoch": 108, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 42228, + "train_examples": 50000, + "train_loss": 0.046444710136651995 + }, + { + "epoch": 109, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 42619, + "train_examples": 50000, + "train_loss": 0.04350686324119568 + }, + { + "epoch": 110, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 43010, + "train_examples": 50000, + "train_loss": 0.03984828278422356 + }, + { + "epoch": 111, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 43401, + "train_examples": 50000, + "train_loss": 0.038114627991318706 + }, + { + "epoch": 112, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 43792, + "train_examples": 50000, + "train_loss": 0.03725605479836464 + }, + { + "epoch": 113, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 44183, + "train_examples": 50000, + "train_loss": 0.035026667524576185 + }, + { + "epoch": 114, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 44574, + "train_examples": 50000, + "train_loss": 0.03393690106987953 + }, + { + "epoch": 115, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 44965, + "train_examples": 50000, + "train_loss": 0.031377520961761475 + }, + { + "epoch": 116, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 45356, + "train_examples": 50000, + "train_loss": 0.03009181555867195 + }, + { + "epoch": 117, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 45747, + "train_examples": 50000, + "train_loss": 0.028700488542318343 + }, + { + "epoch": 118, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 46138, + "train_examples": 50000, + "train_loss": 0.028414438714981078 + }, + { + "epoch": 119, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 46529, + "train_examples": 50000, + "train_loss": 0.02713586230635643 + }, + { + "epoch": 120, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 46920, + "train_examples": 50000, + "train_loss": 0.02669134171426296 + }, + { + "epoch": 121, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 47311, + "train_examples": 50000, + "train_loss": 0.024967432166934013 + }, + { + "epoch": 122, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 47702, + "train_examples": 50000, + "train_loss": 0.023414477905631066 + }, + { + "epoch": 123, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 48093, + "train_examples": 50000, + "train_loss": 0.024176200451254844 + }, + { + "epoch": 124, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 48484, + "train_examples": 50000, + "train_loss": 0.02272428689658642 + }, + { + "epoch": 125, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 48875, + "train_examples": 50000, + "train_loss": 0.02198539039373398 + }, + { + "epoch": 126, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 49266, + "train_examples": 50000, + "train_loss": 0.02231282145857811 + }, + { + "epoch": 127, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 49657, + "train_examples": 50000, + "train_loss": 0.020591423956751825 + }, + { + "epoch": 128, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 50048, + "train_examples": 50000, + "train_loss": 0.020472578698694704 + }, + { + "epoch": 129, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 50439, + "train_examples": 50000, + "train_loss": 0.02021690424859524 + }, + { + "epoch": 130, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 50830, + "train_examples": 50000, + "train_loss": 0.019907781376838685 + }, + { + "epoch": 131, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 51221, + "train_examples": 50000, + "train_loss": 0.01926221474170685 + }, + { + "epoch": 132, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 51612, + "train_examples": 50000, + "train_loss": 0.01843577579677105 + }, + { + "epoch": 133, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 52003, + "train_examples": 50000, + "train_loss": 0.017993157770186664 + }, + { + "epoch": 134, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 52394, + "train_examples": 50000, + "train_loss": 0.01773969442784786 + }, + { + "epoch": 135, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 52785, + "train_examples": 50000, + "train_loss": 0.016086643611192702 + }, + { + "epoch": 136, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 53176, + "train_examples": 50000, + "train_loss": 0.015963542224168778 + }, + { + "epoch": 137, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 53567, + "train_examples": 50000, + "train_loss": 0.01700747108578682 + }, + { + "epoch": 138, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 53958, + "train_examples": 50000, + "train_loss": 0.01534853366971016 + }, + { + "epoch": 139, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 54349, + "train_examples": 50000, + "train_loss": 0.016440005310773848 + }, + { + "epoch": 140, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 54740, + "train_examples": 50000, + "train_loss": 0.01649655404508114 + }, + { + "epoch": 141, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 55131, + "train_examples": 50000, + "train_loss": 0.015004711900502444 + }, + { + "epoch": 142, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 55522, + "train_examples": 50000, + "train_loss": 0.015482185236811639 + }, + { + "epoch": 143, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 55913, + "train_examples": 50000, + "train_loss": 0.01354793422177434 + }, + { + "epoch": 144, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 56304, + "train_examples": 50000, + "train_loss": 0.013621482223570347 + }, + { + "epoch": 145, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 56695, + "train_examples": 50000, + "train_loss": 0.013704320205636323 + }, + { + "epoch": 146, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 57086, + "train_examples": 50000, + "train_loss": 0.01413012717962265 + }, + { + "epoch": 147, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 57477, + "train_examples": 50000, + "train_loss": 0.014088259098529815 + }, + { + "epoch": 148, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 57868, + "train_examples": 50000, + "train_loss": 0.012782217108905316 + }, + { + "epoch": 149, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 58259, + "train_examples": 50000, + "train_loss": 0.012891866796314716 + }, + { + "epoch": 150, + "lr": 0.010000000000000002, + "output_lr": 0.010000000000000002, + "step": 58650, + "train_examples": 50000, + "train_loss": 0.014231026040911674 + }, + { + "epoch": 151, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 59041, + "train_examples": 50000, + "train_loss": 0.012376270772218704 + }, + { + "epoch": 152, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 59432, + "train_examples": 50000, + "train_loss": 0.010853191240429879 + }, + { + "epoch": 153, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 59823, + "train_examples": 50000, + "train_loss": 0.010597322267368436 + }, + { + "epoch": 154, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 60214, + "train_examples": 50000, + "train_loss": 0.010722288198769093 + }, + { + "epoch": 155, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 60605, + "train_examples": 50000, + "train_loss": 0.010636799686178564 + }, + { + "epoch": 156, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 60996, + "train_examples": 50000, + "train_loss": 0.010403271390199661 + }, + { + "epoch": 157, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 61387, + "train_examples": 50000, + "train_loss": 0.009927612109184265 + }, + { + "epoch": 158, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 61778, + "train_examples": 50000, + "train_loss": 0.009928871421813965 + }, + { + "epoch": 159, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 62169, + "train_examples": 50000, + "train_loss": 0.010260844878703356 + }, + { + "epoch": 160, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 62560, + "train_examples": 50000, + "train_loss": 0.009526269187107682 + }, + { + "epoch": 161, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 62951, + "train_examples": 50000, + "train_loss": 0.009347029538601637 + }, + { + "epoch": 162, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 63342, + "train_examples": 50000, + "train_loss": 0.009664224430173635 + }, + { + "epoch": 163, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 63733, + "train_examples": 50000, + "train_loss": 0.009358572429418564 + }, + { + "epoch": 164, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 64124, + "train_examples": 50000, + "train_loss": 0.008715892453491688 + }, + { + "epoch": 165, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 64515, + "train_examples": 50000, + "train_loss": 0.009265132047608494 + }, + { + "epoch": 166, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 64906, + "train_examples": 50000, + "train_loss": 0.009125266028791665 + }, + { + "epoch": 167, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 65297, + "train_examples": 50000, + "train_loss": 0.009044252778589725 + }, + { + "epoch": 168, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 65688, + "train_examples": 50000, + "train_loss": 0.008436540015377103 + }, + { + "epoch": 169, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 66079, + "train_examples": 50000, + "train_loss": 0.00821531246304512 + }, + { + "epoch": 170, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 66470, + "train_examples": 50000, + "train_loss": 0.009059850264191628 + }, + { + "epoch": 171, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 66861, + "train_examples": 50000, + "train_loss": 0.00861273598548025 + }, + { + "epoch": 172, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 67252, + "train_examples": 50000, + "train_loss": 0.008712154743671418 + }, + { + "epoch": 173, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 67643, + "train_examples": 50000, + "train_loss": 0.009049842408746482 + }, + { + "epoch": 174, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 68034, + "train_examples": 50000, + "train_loss": 0.009108233963102102 + }, + { + "epoch": 175, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 68425, + "train_examples": 50000, + "train_loss": 0.0075224792397022245 + }, + { + "epoch": 176, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 68816, + "train_examples": 50000, + "train_loss": 0.009295342291593551 + }, + { + "epoch": 177, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 69207, + "train_examples": 50000, + "train_loss": 0.008169813227802514 + }, + { + "epoch": 178, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 69598, + "train_examples": 50000, + "train_loss": 0.00769507429793477 + }, + { + "epoch": 179, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 69989, + "train_examples": 50000, + "train_loss": 0.008760777043998242 + }, + { + "epoch": 180, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 70380, + "train_examples": 50000, + "train_loss": 0.00868443419508636 + }, + { + "epoch": 181, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 70771, + "train_examples": 50000, + "train_loss": 0.008081732947379352 + }, + { + "epoch": 182, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 71162, + "train_examples": 50000, + "train_loss": 0.008519896957427263 + }, + { + "epoch": 183, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 71553, + "train_examples": 50000, + "train_loss": 0.008458584978729486 + }, + { + "epoch": 184, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 71944, + "train_examples": 50000, + "train_loss": 0.008384161400943994 + }, + { + "epoch": 185, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 72335, + "train_examples": 50000, + "train_loss": 0.008155007639229297 + }, + { + "epoch": 186, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 72726, + "train_examples": 50000, + "train_loss": 0.008052484165802598 + }, + { + "epoch": 187, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 73117, + "train_examples": 50000, + "train_loss": 0.008273763872981071 + }, + { + "epoch": 188, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 73508, + "train_examples": 50000, + "train_loss": 0.007871800690144301 + }, + { + "epoch": 189, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 73899, + "train_examples": 50000, + "train_loss": 0.008268070443421602 + }, + { + "epoch": 190, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 74290, + "train_examples": 50000, + "train_loss": 0.00846611877180636 + }, + { + "epoch": 191, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 74681, + "train_examples": 50000, + "train_loss": 0.00802742943264544 + }, + { + "epoch": 192, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 75072, + "train_examples": 50000, + "train_loss": 0.008076108279377221 + }, + { + "epoch": 193, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 75463, + "train_examples": 50000, + "train_loss": 0.007987720090746879 + }, + { + "epoch": 194, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 75854, + "train_examples": 50000, + "train_loss": 0.008275099214911461 + }, + { + "epoch": 195, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 76245, + "train_examples": 50000, + "train_loss": 0.007724166117459535 + }, + { + "epoch": 196, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 76636, + "train_examples": 50000, + "train_loss": 0.00832881974235177 + }, + { + "epoch": 197, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 77027, + "train_examples": 50000, + "train_loss": 0.007771533735245466 + }, + { + "epoch": 198, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 77418, + "train_examples": 50000, + "train_loss": 0.008149257074594497 + }, + { + "epoch": 199, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 77809, + "train_examples": 50000, + "train_loss": 0.008355577214360237 + }, + { + "epoch": 200, + "lr": 0.0010000000000000002, + "output_lr": 0.0010000000000000002, + "step": 78200, + "train_examples": 50000, + "train_loss": 0.007793944703489542 + } + ], + "evaluation_protocol": { + "test_evaluations": 1, + "test_used_for_selection": false, + "validation_evaluations": 0 + }, + "final": { + "accuracy": 0.9179, + "epoch": 200, + "evaluation_split": "test", + "finite": true, + "loss": 0.41951209106445314, + "step": 78200 + }, + "hardware": { + "cuda_device_name": "NVIDIA GeForce GTX 1080", + "cuda_visible_devices": "5", + "device": "cuda", + "device_total_memory_bytes": 8507949056, + "peak_memory_allocated_bytes": 2530507264, + "peak_memory_reserved_bytes": 3254779904, + "torch_version": "2.3.1+cu118" + }, + "protocol_family": "oral_a_cifar_local_resnet_development", + "provenance": { + "git_commit": "8f9364e8ee2e2a19eb83b58452fa0e8cb155353c", + "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": "test", + "input_layout": "NCHW", + "input_shape": [ + 3, + 32, + 32 + ], + "loader_seed": 10, + "normalization_mean": [ + 0.49140000343322754, + 0.4821999967098236, + 0.4465000033378601 + ], + "normalization_std": [ + 0.24699999392032623, + 0.2434999942779541, + 0.26159998774528503 + ], + "split_from_training_only": true, + "split_seed": null, + "test_examples": 10000, + "train_examples": 50000, + "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5", + "validation_class_counts": {}, + "validation_examples": 0, + "validation_index_sha256": null + }, + "timing": { + "apical_warmup_wall_s": 0.0, + "evaluation_wall_s": 1.1608846187591553, + "mirror_warmup_wall_s": 0.0, + "predictor_warmup_wall_s": 0.0, + "timing_excludes_data_loading_hashing_and_model_construction": true, + "total_timed_wall_s": 5888.902309417725, + "train_wall_s": 5887.7052166461945 + }, + "work": { + "apical_macs_per_example": 0, + "causal_scalar_observations": 0, + "components": { + "apical_projection_macs": 0, + "apical_regression_macs": 0, + "bp_reverse_macs_estimate": 1377251840000000, + "kp_reciprocal_correlation_macs": 0, + "local_weight_correlation_macs": 0, + "mirror_feedback_prediction_macs": 0, + "mirror_local_correlation_macs": 0, + "mirror_response_macs": 0, + "ordinary_forward_macs": 688625920000000, + "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; mixed-traffic/predictor elementwise arithmetic is reported as a conservative operation estimate separately and is not folded into MACs", + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 68862592, + "logical_batch_loss_queries": 0, + "mirror_probe_examples": 0, + "mirror_readout_probe_examples": 0, + "neutral_projection_observations": 0, + "per_example_cross_entropy_terms": 0, + "total_clean_forward_examples": 10000000, + "total_forward_equivalent_examples": 10000000, + "total_macs_estimate": 2065877760000000 + } +} |
