summaryrefslogtreecommitdiff
path: root/results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json
diff options
context:
space:
mode:
Diffstat (limited to 'results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json')
-rw-r--r--results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json2089
1 files changed, 2089 insertions, 0 deletions
diff --git a/results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json b/results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json
new file mode 100644
index 0000000..cb2b886
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json
@@ -0,0 +1,2089 @@
+{
+ "architecture": {
+ "adaptive_apical_parameters": 3637248,
+ "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": 303104,
+ "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": 606208,
+ "residual_scale": 1.0,
+ "vectorizer_mode": "spatial_template",
+ "vectorizer_parameters": 3031040
+ },
+ "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": 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": 0.03,
+ "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": "dfa",
+ "momentum": 0.9,
+ "neutral_projection": 0,
+ "normalization": "batchnorm",
+ "nuisance_scale": 0.0,
+ "out": "results/oral_a_dynamic_scaling_v2/dfa_d32_s10.json",
+ "output_lr": 0.1,
+ "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": "full_hidden_field",
+ "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": {
+ "early_third_mean": 3.074540582019836e-05,
+ "innovation_negative_gradient_cosine": [
+ -0.00036826980067417026,
+ 2.6044523110613227e-05,
+ 0.0008353812154382467,
+ -0.0008176610572263598,
+ 0.0013316046679392457,
+ -0.0018267796840518713,
+ -0.00035030735307373106,
+ 0.00024559281882829964,
+ -0.0017333189025521278,
+ 0.0029651676304638386,
+ -0.0007047887775115669,
+ 0.0012012019287794828,
+ -0.0003103498602285981,
+ 0.0016303639858961105,
+ 0.0007953367312438786,
+ 0.002465467667207122,
+ 0.0007311954977922142,
+ -0.003911031410098076,
+ 0.0066788699477910995,
+ 0.0017323954962193966,
+ 0.001733468729071319,
+ 0.007426526863127947,
+ 0.009960150346159935,
+ 0.009964510798454285,
+ 0.013116281479597092,
+ 0.00614601606503129,
+ 0.013978250324726105,
+ 0.008247887715697289,
+ 0.015481583774089813,
+ 0.013867078348994255,
+ 0.01987851783633232
+ ],
+ "normalization_state": "training_batch_stats_without_running_update",
+ "raw_negative_gradient_cosine": [
+ -0.00036826980067417026,
+ 2.6044523110613227e-05,
+ 0.0008353812154382467,
+ -0.0008176610572263598,
+ 0.0013316046679392457,
+ -0.0018267796840518713,
+ -0.00035030735307373106,
+ 0.00024559281882829964,
+ -0.0017333189025521278,
+ 0.0029651676304638386,
+ -0.0007047887775115669,
+ 0.0012012019287794828,
+ -0.0003103498602285981,
+ 0.0016303639858961105,
+ 0.0007953367312438786,
+ 0.002465467667207122,
+ 0.0007311954977922142,
+ -0.003911031410098076,
+ 0.0066788699477910995,
+ 0.0017323954962193966,
+ 0.001733468729071319,
+ 0.007426526863127947,
+ 0.009960150346159935,
+ 0.009964510798454285,
+ 0.013116281479597092,
+ 0.00614601606503129,
+ 0.013978250324726105,
+ 0.008247887715697289,
+ 0.015481583774089813,
+ 0.013867078348994255,
+ 0.01987851783633232
+ ],
+ "teaching_negative_gradient_cosine": [
+ -0.00036826980067417026,
+ 2.6044523110613227e-05,
+ 0.0008353812154382467,
+ -0.0008176610572263598,
+ 0.0013316046679392457,
+ -0.0018267796840518713,
+ -0.00035030735307373106,
+ 0.00024559281882829964,
+ -0.0017333189025521278,
+ 0.0029651676304638386,
+ -0.0007047887775115669,
+ 0.0012012019287794828,
+ -0.0003103498602285981,
+ 0.0016303639858961105,
+ 0.0007953367312438786,
+ 0.002465467667207122,
+ 0.0007311954977922142,
+ -0.003911031410098076,
+ 0.0066788699477910995,
+ 0.0017323954962193966,
+ 0.001733468729071319,
+ 0.007426526863127947,
+ 0.009960150346159935,
+ 0.009964510798454285,
+ 0.013116281479597092,
+ 0.00614601606503129,
+ 0.013978250324726105,
+ 0.008247887715697289,
+ 0.015481583774089813,
+ 0.013867078348994255,
+ 0.01987851783633232
+ ],
+ "wall_s": 0.14246678352355957
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 391,
+ "train_examples": 50000,
+ "train_loss": 6.293598889007568
+ },
+ {
+ "epoch": 2,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 782,
+ "train_examples": 50000,
+ "train_loss": 7.412033297271728
+ },
+ {
+ "epoch": 3,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 1173,
+ "train_examples": 50000,
+ "train_loss": 10.737898612060548
+ },
+ {
+ "epoch": 4,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 1564,
+ "train_examples": 50000,
+ "train_loss": 16.133542594604492
+ },
+ {
+ "epoch": 5,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 1955,
+ "train_examples": 50000,
+ "train_loss": 22.19608327026367
+ },
+ {
+ "epoch": 6,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 2346,
+ "train_examples": 50000,
+ "train_loss": 30.915053834228516
+ },
+ {
+ "epoch": 7,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 2737,
+ "train_examples": 50000,
+ "train_loss": 39.64462054199219
+ },
+ {
+ "epoch": 8,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 3128,
+ "train_examples": 50000,
+ "train_loss": 57.555933121337894
+ },
+ {
+ "epoch": 9,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 3519,
+ "train_examples": 50000,
+ "train_loss": 74.06779753051758
+ },
+ {
+ "epoch": 10,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 3910,
+ "train_examples": 50000,
+ "train_loss": 94.07128509277344
+ },
+ {
+ "epoch": 11,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 4301,
+ "train_examples": 50000,
+ "train_loss": 111.10326148925782
+ },
+ {
+ "epoch": 12,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 4692,
+ "train_examples": 50000,
+ "train_loss": 141.19566643066406
+ },
+ {
+ "epoch": 13,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 5083,
+ "train_examples": 50000,
+ "train_loss": 170.659443046875
+ },
+ {
+ "epoch": 14,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 5474,
+ "train_examples": 50000,
+ "train_loss": 194.02587569335938
+ },
+ {
+ "epoch": 15,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 5865,
+ "train_examples": 50000,
+ "train_loss": 233.4889348339844
+ },
+ {
+ "epoch": 16,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 6256,
+ "train_examples": 50000,
+ "train_loss": 271.0032402294922
+ },
+ {
+ "epoch": 17,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 6647,
+ "train_examples": 50000,
+ "train_loss": 314.8979043945312
+ },
+ {
+ "epoch": 18,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 7038,
+ "train_examples": 50000,
+ "train_loss": 347.94102638671876
+ },
+ {
+ "epoch": 19,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 7429,
+ "train_examples": 50000,
+ "train_loss": 409.38084181640625
+ },
+ {
+ "epoch": 20,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 7820,
+ "train_examples": 50000,
+ "train_loss": 441.446261484375
+ },
+ {
+ "epoch": 21,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 8211,
+ "train_examples": 50000,
+ "train_loss": 516.6539449804687
+ },
+ {
+ "epoch": 22,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 8602,
+ "train_examples": 50000,
+ "train_loss": 539.8580530175781
+ },
+ {
+ "epoch": 23,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 8993,
+ "train_examples": 50000,
+ "train_loss": 580.1104050195313
+ },
+ {
+ "epoch": 24,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 9384,
+ "train_examples": 50000,
+ "train_loss": 676.0906620703125
+ },
+ {
+ "epoch": 25,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 9775,
+ "train_examples": 50000,
+ "train_loss": 765.2543776171875
+ },
+ {
+ "epoch": 26,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 10166,
+ "train_examples": 50000,
+ "train_loss": 865.8800613671875
+ },
+ {
+ "epoch": 27,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 10557,
+ "train_examples": 50000,
+ "train_loss": 933.8132362890625
+ },
+ {
+ "epoch": 28,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 10948,
+ "train_examples": 50000,
+ "train_loss": 1060.9912740625
+ },
+ {
+ "epoch": 29,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 11339,
+ "train_examples": 50000,
+ "train_loss": 978.6808122265625
+ },
+ {
+ "epoch": 30,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 11730,
+ "train_examples": 50000,
+ "train_loss": 1180.31563515625
+ },
+ {
+ "epoch": 31,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 12121,
+ "train_examples": 50000,
+ "train_loss": 1244.2332968359376
+ },
+ {
+ "epoch": 32,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 12512,
+ "train_examples": 50000,
+ "train_loss": 1351.088999765625
+ },
+ {
+ "epoch": 33,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 12903,
+ "train_examples": 50000,
+ "train_loss": 1416.4225203125
+ },
+ {
+ "epoch": 34,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 13294,
+ "train_examples": 50000,
+ "train_loss": 1457.549191484375
+ },
+ {
+ "epoch": 35,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 13685,
+ "train_examples": 50000,
+ "train_loss": 1664.370351171875
+ },
+ {
+ "epoch": 36,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 14076,
+ "train_examples": 50000,
+ "train_loss": 1552.4956359375
+ },
+ {
+ "epoch": 37,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 14467,
+ "train_examples": 50000,
+ "train_loss": 1821.0322580078125
+ },
+ {
+ "epoch": 38,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 14858,
+ "train_examples": 50000,
+ "train_loss": 1990.8570221875
+ },
+ {
+ "epoch": 39,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 15249,
+ "train_examples": 50000,
+ "train_loss": 1969.9385286328124
+ },
+ {
+ "epoch": 40,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 15640,
+ "train_examples": 50000,
+ "train_loss": 2325.989310234375
+ },
+ {
+ "epoch": 41,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 16031,
+ "train_examples": 50000,
+ "train_loss": 2438.87965421875
+ },
+ {
+ "epoch": 42,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 16422,
+ "train_examples": 50000,
+ "train_loss": 2770.6505803125
+ },
+ {
+ "epoch": 43,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 16813,
+ "train_examples": 50000,
+ "train_loss": 2711.5188660546874
+ },
+ {
+ "epoch": 44,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 17204,
+ "train_examples": 50000,
+ "train_loss": 2650.04287421875
+ },
+ {
+ "epoch": 45,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 17595,
+ "train_examples": 50000,
+ "train_loss": 2918.512974765625
+ },
+ {
+ "epoch": 46,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 17986,
+ "train_examples": 50000,
+ "train_loss": 3005.5448178125
+ },
+ {
+ "epoch": 47,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 18377,
+ "train_examples": 50000,
+ "train_loss": 3492.73231265625
+ },
+ {
+ "epoch": 48,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 18768,
+ "train_examples": 50000,
+ "train_loss": 3662.0783178125
+ },
+ {
+ "epoch": 49,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 19159,
+ "train_examples": 50000,
+ "train_loss": 3463.054914609375
+ },
+ {
+ "epoch": 50,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 19550,
+ "train_examples": 50000,
+ "train_loss": 3419.4710433984374
+ },
+ {
+ "epoch": 51,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 19941,
+ "train_examples": 50000,
+ "train_loss": 3948.77810875
+ },
+ {
+ "epoch": 52,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 20332,
+ "train_examples": 50000,
+ "train_loss": 4149.43587421875
+ },
+ {
+ "epoch": 53,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 20723,
+ "train_examples": 50000,
+ "train_loss": 4289.252799921875
+ },
+ {
+ "epoch": 54,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 21114,
+ "train_examples": 50000,
+ "train_loss": 4122.716543242187
+ },
+ {
+ "epoch": 55,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 21505,
+ "train_examples": 50000,
+ "train_loss": 4666.4023015625
+ },
+ {
+ "epoch": 56,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 21896,
+ "train_examples": 50000,
+ "train_loss": 4463.53142515625
+ },
+ {
+ "epoch": 57,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 22287,
+ "train_examples": 50000,
+ "train_loss": 4326.788139375
+ },
+ {
+ "epoch": 58,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 22678,
+ "train_examples": 50000,
+ "train_loss": 5110.3531796875
+ },
+ {
+ "epoch": 59,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 23069,
+ "train_examples": 50000,
+ "train_loss": 5457.86820171875
+ },
+ {
+ "epoch": 60,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 23460,
+ "train_examples": 50000,
+ "train_loss": 5612.76991125
+ },
+ {
+ "epoch": 61,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 23851,
+ "train_examples": 50000,
+ "train_loss": 5747.9700634375
+ },
+ {
+ "epoch": 62,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 24242,
+ "train_examples": 50000,
+ "train_loss": 5802.20757921875
+ },
+ {
+ "epoch": 63,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 24633,
+ "train_examples": 50000,
+ "train_loss": 5967.9336390625
+ },
+ {
+ "epoch": 64,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 25024,
+ "train_examples": 50000,
+ "train_loss": 6246.990884921875
+ },
+ {
+ "epoch": 65,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 25415,
+ "train_examples": 50000,
+ "train_loss": 6950.7405046875
+ },
+ {
+ "epoch": 66,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 25806,
+ "train_examples": 50000,
+ "train_loss": 6945.5849509375
+ },
+ {
+ "epoch": 67,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 26197,
+ "train_examples": 50000,
+ "train_loss": 6585.1524090625
+ },
+ {
+ "epoch": 68,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 26588,
+ "train_examples": 50000,
+ "train_loss": 6822.074729375
+ },
+ {
+ "epoch": 69,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 26979,
+ "train_examples": 50000,
+ "train_loss": 7025.1091890625
+ },
+ {
+ "epoch": 70,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 27370,
+ "train_examples": 50000,
+ "train_loss": 8168.0862190625
+ },
+ {
+ "epoch": 71,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 27761,
+ "train_examples": 50000,
+ "train_loss": 8214.2977309375
+ },
+ {
+ "epoch": 72,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 28152,
+ "train_examples": 50000,
+ "train_loss": 7491.4265615625
+ },
+ {
+ "epoch": 73,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 28543,
+ "train_examples": 50000,
+ "train_loss": 8174.256045625
+ },
+ {
+ "epoch": 74,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 28934,
+ "train_examples": 50000,
+ "train_loss": 8563.91816875
+ },
+ {
+ "epoch": 75,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 29325,
+ "train_examples": 50000,
+ "train_loss": 8911.6764040625
+ },
+ {
+ "epoch": 76,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 29716,
+ "train_examples": 50000,
+ "train_loss": 8645.0338546875
+ },
+ {
+ "epoch": 77,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 30107,
+ "train_examples": 50000,
+ "train_loss": 8843.74548875
+ },
+ {
+ "epoch": 78,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 30498,
+ "train_examples": 50000,
+ "train_loss": 9447.9077440625
+ },
+ {
+ "epoch": 79,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 30889,
+ "train_examples": 50000,
+ "train_loss": 9591.966471875
+ },
+ {
+ "epoch": 80,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 31280,
+ "train_examples": 50000,
+ "train_loss": 9565.742981875
+ },
+ {
+ "epoch": 81,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 31671,
+ "train_examples": 50000,
+ "train_loss": 10660.2193578125
+ },
+ {
+ "epoch": 82,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 32062,
+ "train_examples": 50000,
+ "train_loss": 11309.25382125
+ },
+ {
+ "epoch": 83,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 32453,
+ "train_examples": 50000,
+ "train_loss": 10361.5855065625
+ },
+ {
+ "epoch": 84,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 32844,
+ "train_examples": 50000,
+ "train_loss": 10960.4349828125
+ },
+ {
+ "epoch": 85,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 33235,
+ "train_examples": 50000,
+ "train_loss": 11115.4821184375
+ },
+ {
+ "epoch": 86,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 33626,
+ "train_examples": 50000,
+ "train_loss": 11953.2073675
+ },
+ {
+ "epoch": 87,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 34017,
+ "train_examples": 50000,
+ "train_loss": 11733.153768125
+ },
+ {
+ "epoch": 88,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 34408,
+ "train_examples": 50000,
+ "train_loss": 11426.63758875
+ },
+ {
+ "epoch": 89,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 34799,
+ "train_examples": 50000,
+ "train_loss": 13487.809304375
+ },
+ {
+ "epoch": 90,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 35190,
+ "train_examples": 50000,
+ "train_loss": 13246.3208703125
+ },
+ {
+ "epoch": 91,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 35581,
+ "train_examples": 50000,
+ "train_loss": 12576.546934375
+ },
+ {
+ "epoch": 92,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 35972,
+ "train_examples": 50000,
+ "train_loss": 14237.348911875
+ },
+ {
+ "epoch": 93,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 36363,
+ "train_examples": 50000,
+ "train_loss": 13395.600990625
+ },
+ {
+ "epoch": 94,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 36754,
+ "train_examples": 50000,
+ "train_loss": 15017.1467109375
+ },
+ {
+ "epoch": 95,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 37145,
+ "train_examples": 50000,
+ "train_loss": 14967.916365625
+ },
+ {
+ "epoch": 96,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 37536,
+ "train_examples": 50000,
+ "train_loss": 14193.336595625
+ },
+ {
+ "epoch": 97,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 37927,
+ "train_examples": 50000,
+ "train_loss": 13713.72429125
+ },
+ {
+ "epoch": 98,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 38318,
+ "train_examples": 50000,
+ "train_loss": 16357.4106659375
+ },
+ {
+ "epoch": 99,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 38709,
+ "train_examples": 50000,
+ "train_loss": 14531.0011725
+ },
+ {
+ "epoch": 100,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 39100,
+ "train_examples": 50000,
+ "train_loss": 15943.223449375
+ },
+ {
+ "epoch": 101,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 39491,
+ "train_examples": 50000,
+ "train_loss": 3344.21362453125
+ },
+ {
+ "epoch": 102,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 39882,
+ "train_examples": 50000,
+ "train_loss": 2304.325952890625
+ },
+ {
+ "epoch": 103,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 40273,
+ "train_examples": 50000,
+ "train_loss": 2105.208774375
+ },
+ {
+ "epoch": 104,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 40664,
+ "train_examples": 50000,
+ "train_loss": 2003.5612227734375
+ },
+ {
+ "epoch": 105,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 41055,
+ "train_examples": 50000,
+ "train_loss": 1844.6388519140626
+ },
+ {
+ "epoch": 106,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 41446,
+ "train_examples": 50000,
+ "train_loss": 1716.0516268359374
+ },
+ {
+ "epoch": 107,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 41837,
+ "train_examples": 50000,
+ "train_loss": 1653.430671640625
+ },
+ {
+ "epoch": 108,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 42228,
+ "train_examples": 50000,
+ "train_loss": 1530.946092265625
+ },
+ {
+ "epoch": 109,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 42619,
+ "train_examples": 50000,
+ "train_loss": 1437.4951969140625
+ },
+ {
+ "epoch": 110,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 43010,
+ "train_examples": 50000,
+ "train_loss": 1405.1221304296876
+ },
+ {
+ "epoch": 111,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 43401,
+ "train_examples": 50000,
+ "train_loss": 1343.4887116015625
+ },
+ {
+ "epoch": 112,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 43792,
+ "train_examples": 50000,
+ "train_loss": 1290.05945640625
+ },
+ {
+ "epoch": 113,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 44183,
+ "train_examples": 50000,
+ "train_loss": 1261.2783257421875
+ },
+ {
+ "epoch": 114,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 44574,
+ "train_examples": 50000,
+ "train_loss": 1188.335002421875
+ },
+ {
+ "epoch": 115,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 44965,
+ "train_examples": 50000,
+ "train_loss": 1244.6512447265625
+ },
+ {
+ "epoch": 116,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 45356,
+ "train_examples": 50000,
+ "train_loss": 1145.0465993359376
+ },
+ {
+ "epoch": 117,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 45747,
+ "train_examples": 50000,
+ "train_loss": 1258.005577734375
+ },
+ {
+ "epoch": 118,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 46138,
+ "train_examples": 50000,
+ "train_loss": 1194.1361692578125
+ },
+ {
+ "epoch": 119,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 46529,
+ "train_examples": 50000,
+ "train_loss": 1178.327138203125
+ },
+ {
+ "epoch": 120,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 46920,
+ "train_examples": 50000,
+ "train_loss": 1151.9809562695314
+ },
+ {
+ "epoch": 121,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 47311,
+ "train_examples": 50000,
+ "train_loss": 1115.4522632421874
+ },
+ {
+ "epoch": 122,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 47702,
+ "train_examples": 50000,
+ "train_loss": 1212.5351327734375
+ },
+ {
+ "epoch": 123,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 48093,
+ "train_examples": 50000,
+ "train_loss": 1236.7264633203124
+ },
+ {
+ "epoch": 124,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 48484,
+ "train_examples": 50000,
+ "train_loss": 1129.7096988671874
+ },
+ {
+ "epoch": 125,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 48875,
+ "train_examples": 50000,
+ "train_loss": 1124.5999420703124
+ },
+ {
+ "epoch": 126,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 49266,
+ "train_examples": 50000,
+ "train_loss": 1214.3487353125
+ },
+ {
+ "epoch": 127,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 49657,
+ "train_examples": 50000,
+ "train_loss": 1097.898828359375
+ },
+ {
+ "epoch": 128,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 50048,
+ "train_examples": 50000,
+ "train_loss": 1076.2484835546875
+ },
+ {
+ "epoch": 129,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 50439,
+ "train_examples": 50000,
+ "train_loss": 1247.9283218359376
+ },
+ {
+ "epoch": 130,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 50830,
+ "train_examples": 50000,
+ "train_loss": 1214.075158125
+ },
+ {
+ "epoch": 131,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 51221,
+ "train_examples": 50000,
+ "train_loss": 1198.93064765625
+ },
+ {
+ "epoch": 132,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 51612,
+ "train_examples": 50000,
+ "train_loss": 1142.9898784375
+ },
+ {
+ "epoch": 133,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 52003,
+ "train_examples": 50000,
+ "train_loss": 1205.919777890625
+ },
+ {
+ "epoch": 134,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 52394,
+ "train_examples": 50000,
+ "train_loss": 1262.817801171875
+ },
+ {
+ "epoch": 135,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 52785,
+ "train_examples": 50000,
+ "train_loss": 1210.975372265625
+ },
+ {
+ "epoch": 136,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 53176,
+ "train_examples": 50000,
+ "train_loss": 1231.7331519726563
+ },
+ {
+ "epoch": 137,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 53567,
+ "train_examples": 50000,
+ "train_loss": 1328.930333359375
+ },
+ {
+ "epoch": 138,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 53958,
+ "train_examples": 50000,
+ "train_loss": 1254.761289140625
+ },
+ {
+ "epoch": 139,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 54349,
+ "train_examples": 50000,
+ "train_loss": 1125.5186128125
+ },
+ {
+ "epoch": 140,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 54740,
+ "train_examples": 50000,
+ "train_loss": 1262.8685775195313
+ },
+ {
+ "epoch": 141,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 55131,
+ "train_examples": 50000,
+ "train_loss": 1237.459378671875
+ },
+ {
+ "epoch": 142,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 55522,
+ "train_examples": 50000,
+ "train_loss": 1204.0073876953124
+ },
+ {
+ "epoch": 143,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 55913,
+ "train_examples": 50000,
+ "train_loss": 1137.405095078125
+ },
+ {
+ "epoch": 144,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 56304,
+ "train_examples": 50000,
+ "train_loss": 1258.9069240625
+ },
+ {
+ "epoch": 145,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 56695,
+ "train_examples": 50000,
+ "train_loss": 1275.553220390625
+ },
+ {
+ "epoch": 146,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 57086,
+ "train_examples": 50000,
+ "train_loss": 1346.54895
+ },
+ {
+ "epoch": 147,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 57477,
+ "train_examples": 50000,
+ "train_loss": 1248.5952147265625
+ },
+ {
+ "epoch": 148,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 57868,
+ "train_examples": 50000,
+ "train_loss": 1208.6872989648436
+ },
+ {
+ "epoch": 149,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 58259,
+ "train_examples": 50000,
+ "train_loss": 1295.82844421875
+ },
+ {
+ "epoch": 150,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 58650,
+ "train_examples": 50000,
+ "train_loss": 1316.0796097070313
+ },
+ {
+ "epoch": 151,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 59041,
+ "train_examples": 50000,
+ "train_loss": 432.080549765625
+ },
+ {
+ "epoch": 152,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 59432,
+ "train_examples": 50000,
+ "train_loss": 381.3125614453125
+ },
+ {
+ "epoch": 153,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 59823,
+ "train_examples": 50000,
+ "train_loss": 359.7443784375
+ },
+ {
+ "epoch": 154,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 60214,
+ "train_examples": 50000,
+ "train_loss": 330.96993841796876
+ },
+ {
+ "epoch": 155,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 60605,
+ "train_examples": 50000,
+ "train_loss": 320.38391126953127
+ },
+ {
+ "epoch": 156,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 60996,
+ "train_examples": 50000,
+ "train_loss": 299.92142598632813
+ },
+ {
+ "epoch": 157,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 61387,
+ "train_examples": 50000,
+ "train_loss": 292.0546185205078
+ },
+ {
+ "epoch": 158,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 61778,
+ "train_examples": 50000,
+ "train_loss": 283.10320901367186
+ },
+ {
+ "epoch": 159,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 62169,
+ "train_examples": 50000,
+ "train_loss": 273.0513530078125
+ },
+ {
+ "epoch": 160,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 62560,
+ "train_examples": 50000,
+ "train_loss": 267.9574522070312
+ },
+ {
+ "epoch": 161,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 62951,
+ "train_examples": 50000,
+ "train_loss": 262.1898098388672
+ },
+ {
+ "epoch": 162,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 63342,
+ "train_examples": 50000,
+ "train_loss": 251.405320703125
+ },
+ {
+ "epoch": 163,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 63733,
+ "train_examples": 50000,
+ "train_loss": 245.4019890673828
+ },
+ {
+ "epoch": 164,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 64124,
+ "train_examples": 50000,
+ "train_loss": 240.36019327148438
+ },
+ {
+ "epoch": 165,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 64515,
+ "train_examples": 50000,
+ "train_loss": 235.46463319824218
+ },
+ {
+ "epoch": 166,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 64906,
+ "train_examples": 50000,
+ "train_loss": 230.54692152832033
+ },
+ {
+ "epoch": 167,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 65297,
+ "train_examples": 50000,
+ "train_loss": 232.6379735986328
+ },
+ {
+ "epoch": 168,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 65688,
+ "train_examples": 50000,
+ "train_loss": 225.51191487304686
+ },
+ {
+ "epoch": 169,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 66079,
+ "train_examples": 50000,
+ "train_loss": 216.49266020507812
+ },
+ {
+ "epoch": 170,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 66470,
+ "train_examples": 50000,
+ "train_loss": 213.13386258789063
+ },
+ {
+ "epoch": 171,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 66861,
+ "train_examples": 50000,
+ "train_loss": 217.53731814941406
+ },
+ {
+ "epoch": 172,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 67252,
+ "train_examples": 50000,
+ "train_loss": 208.73003357421874
+ },
+ {
+ "epoch": 173,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 67643,
+ "train_examples": 50000,
+ "train_loss": 206.97497490234375
+ },
+ {
+ "epoch": 174,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 68034,
+ "train_examples": 50000,
+ "train_loss": 210.77752537109376
+ },
+ {
+ "epoch": 175,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 68425,
+ "train_examples": 50000,
+ "train_loss": 195.20160494140626
+ },
+ {
+ "epoch": 176,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 68816,
+ "train_examples": 50000,
+ "train_loss": 200.7779848046875
+ },
+ {
+ "epoch": 177,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 69207,
+ "train_examples": 50000,
+ "train_loss": 201.7726231640625
+ },
+ {
+ "epoch": 178,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 69598,
+ "train_examples": 50000,
+ "train_loss": 194.9952972314453
+ },
+ {
+ "epoch": 179,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 69989,
+ "train_examples": 50000,
+ "train_loss": 196.31784922851563
+ },
+ {
+ "epoch": 180,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 70380,
+ "train_examples": 50000,
+ "train_loss": 194.3363351171875
+ },
+ {
+ "epoch": 181,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 70771,
+ "train_examples": 50000,
+ "train_loss": 196.57610537597657
+ },
+ {
+ "epoch": 182,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 71162,
+ "train_examples": 50000,
+ "train_loss": 187.25579179199218
+ },
+ {
+ "epoch": 183,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 71553,
+ "train_examples": 50000,
+ "train_loss": 186.91551185546874
+ },
+ {
+ "epoch": 184,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 71944,
+ "train_examples": 50000,
+ "train_loss": 183.84816095214845
+ },
+ {
+ "epoch": 185,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 72335,
+ "train_examples": 50000,
+ "train_loss": 183.88843528320314
+ },
+ {
+ "epoch": 186,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 72726,
+ "train_examples": 50000,
+ "train_loss": 189.78843431640624
+ },
+ {
+ "epoch": 187,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 73117,
+ "train_examples": 50000,
+ "train_loss": 187.7555524658203
+ },
+ {
+ "epoch": 188,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 73508,
+ "train_examples": 50000,
+ "train_loss": 185.55689125976562
+ },
+ {
+ "epoch": 189,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 73899,
+ "train_examples": 50000,
+ "train_loss": 181.3828788183594
+ },
+ {
+ "epoch": 190,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 74290,
+ "train_examples": 50000,
+ "train_loss": 179.399421640625
+ },
+ {
+ "epoch": 191,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 74681,
+ "train_examples": 50000,
+ "train_loss": 179.9919827246094
+ },
+ {
+ "epoch": 192,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 75072,
+ "train_examples": 50000,
+ "train_loss": 176.98932411865235
+ },
+ {
+ "epoch": 193,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 75463,
+ "train_examples": 50000,
+ "train_loss": 180.73735264648437
+ },
+ {
+ "epoch": 194,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 75854,
+ "train_examples": 50000,
+ "train_loss": 183.74677614746093
+ },
+ {
+ "epoch": 195,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 76245,
+ "train_examples": 50000,
+ "train_loss": 179.24562399414063
+ },
+ {
+ "epoch": 196,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 76636,
+ "train_examples": 50000,
+ "train_loss": 185.12150788574218
+ },
+ {
+ "epoch": 197,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 77027,
+ "train_examples": 50000,
+ "train_loss": 179.334324765625
+ },
+ {
+ "epoch": 198,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 77418,
+ "train_examples": 50000,
+ "train_loss": 171.57741375
+ },
+ {
+ "epoch": 199,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 77809,
+ "train_examples": 50000,
+ "train_loss": 180.4202580859375
+ },
+ {
+ "epoch": 200,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 78200,
+ "train_examples": 50000,
+ "train_loss": 174.17015067871094
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 1,
+ "test_used_for_selection": false,
+ "validation_evaluations": 0
+ },
+ "final": {
+ "accuracy": 0.3521,
+ "epoch": 200,
+ "evaluation_split": "test",
+ "finite": true,
+ "loss": 161.696659375,
+ "step": 78200
+ },
+ "hardware": {
+ "cuda_device_name": "NVIDIA GeForce GTX 1080",
+ "cuda_visible_devices": "7",
+ "device": "cuda",
+ "device_total_memory_bytes": 8507949056,
+ "peak_memory_allocated_bytes": 2537912832,
+ "peak_memory_reserved_bytes": 3323985920,
+ "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": 1.146927833557129,
+ "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": 5882.5757439136505,
+ "train_wall_s": 5881.403960943222
+ },
+ "work": {
+ "apical_macs_per_example": 3031040,
+ "causal_scalar_observations": 0,
+ "components": {
+ "apical_projection_macs": 30310400000000,
+ "apical_regression_macs": 0,
+ "bp_reverse_macs_estimate": 0,
+ "kp_reciprocal_correlation_macs": 0,
+ "local_weight_correlation_macs": 688625920000000,
+ "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": 1407562240000000
+ }
+}