summaryrefslogtreecommitdiff
path: root/results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.json
diff options
context:
space:
mode:
Diffstat (limited to 'results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.json')
-rw-r--r--results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.json2537
1 files changed, 2537 insertions, 0 deletions
diff --git a/results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.json b/results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.json
new file mode 100644
index 0000000..548e13e
--- /dev/null
+++ b/results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.json
@@ -0,0 +1,2537 @@
+{
+ "architecture": {
+ "adaptive_apical_parameters": 0,
+ "adaptive_feedback_parameters": 0,
+ "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": 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": "/orion/u/oscarwan/scratch/sdil-data",
+ "depth": 20,
+ "device": "cuda",
+ "epochs": 200,
+ "eta_A": 0.01,
+ "eta_P": 0.01,
+ "eval_every": 1,
+ "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": 0,
+ "mode": "bp",
+ "momentum": 0.9,
+ "neutral_projection": 0,
+ "normalization": "batchnorm",
+ "nuisance_scale": 0.0,
+ "out": "/orion/u/oscarwan/scratch/sdil-work/sdil-a6000-full-rerun/results/resnet_crossover/r2/resnet-r2-bp-d20-lr0p1.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": 0,
+ "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": 5000,
+ "vectorizer_mode": "spatial_template",
+ "warmup_epochs": 5,
+ "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": 9000000,
+ "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,
+ "eval_accuracy": 0.4004,
+ "eval_loss": 1.7710138671875,
+ "eval_split": "validation",
+ "lr": 0.02,
+ "output_lr": 0.02,
+ "step": 352,
+ "train_examples": 45000,
+ "train_loss": 1.6644643401463826
+ },
+ {
+ "epoch": 2,
+ "eval_accuracy": 0.5672,
+ "eval_loss": 1.221260009765625,
+ "eval_split": "validation",
+ "lr": 0.04,
+ "output_lr": 0.04,
+ "step": 704,
+ "train_examples": 45000,
+ "train_loss": 1.2860143685658774
+ },
+ {
+ "epoch": 3,
+ "eval_accuracy": 0.6406,
+ "eval_loss": 1.0877959594726563,
+ "eval_split": "validation",
+ "lr": 0.06000000000000001,
+ "output_lr": 0.06000000000000001,
+ "step": 1056,
+ "train_examples": 45000,
+ "train_loss": 1.02516774397956
+ },
+ {
+ "epoch": 4,
+ "eval_accuracy": 0.6802,
+ "eval_loss": 0.9318223510742187,
+ "eval_split": "validation",
+ "lr": 0.08,
+ "output_lr": 0.08,
+ "step": 1408,
+ "train_examples": 45000,
+ "train_loss": 0.8597116688304477
+ },
+ {
+ "epoch": 5,
+ "eval_accuracy": 0.7032,
+ "eval_loss": 0.9318287963867188,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 1760,
+ "train_examples": 45000,
+ "train_loss": 0.7553892990854051
+ },
+ {
+ "epoch": 6,
+ "eval_accuracy": 0.737,
+ "eval_loss": 0.864343017578125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 2112,
+ "train_examples": 45000,
+ "train_loss": 0.6687519911342197
+ },
+ {
+ "epoch": 7,
+ "eval_accuracy": 0.7616,
+ "eval_loss": 0.7147335205078125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 2464,
+ "train_examples": 45000,
+ "train_loss": 0.6076406918737624
+ },
+ {
+ "epoch": 8,
+ "eval_accuracy": 0.7384,
+ "eval_loss": 0.8417417602539062,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 2816,
+ "train_examples": 45000,
+ "train_loss": 0.5667279717922211
+ },
+ {
+ "epoch": 9,
+ "eval_accuracy": 0.7888,
+ "eval_loss": 0.6749005004882812,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 3168,
+ "train_examples": 45000,
+ "train_loss": 0.5312818857087029
+ },
+ {
+ "epoch": 10,
+ "eval_accuracy": 0.8044,
+ "eval_loss": 0.602718798828125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 3520,
+ "train_examples": 45000,
+ "train_loss": 0.5028076828956604
+ },
+ {
+ "epoch": 11,
+ "eval_accuracy": 0.8276,
+ "eval_loss": 0.5217552856445312,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 3872,
+ "train_examples": 45000,
+ "train_loss": 0.4791169890032874
+ },
+ {
+ "epoch": 12,
+ "eval_accuracy": 0.7946,
+ "eval_loss": 0.63441904296875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 4224,
+ "train_examples": 45000,
+ "train_loss": 0.4623024181683858
+ },
+ {
+ "epoch": 13,
+ "eval_accuracy": 0.6858,
+ "eval_loss": 1.0865167114257812,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 4576,
+ "train_examples": 45000,
+ "train_loss": 0.43909971352683175
+ },
+ {
+ "epoch": 14,
+ "eval_accuracy": 0.8122,
+ "eval_loss": 0.5628994506835937,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 4928,
+ "train_examples": 45000,
+ "train_loss": 0.42348679066763983
+ },
+ {
+ "epoch": 15,
+ "eval_accuracy": 0.8034,
+ "eval_loss": 0.6175327880859375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 5280,
+ "train_examples": 45000,
+ "train_loss": 0.41002807636260985
+ },
+ {
+ "epoch": 16,
+ "eval_accuracy": 0.8388,
+ "eval_loss": 0.50029931640625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 5632,
+ "train_examples": 45000,
+ "train_loss": 0.39710193020502726
+ },
+ {
+ "epoch": 17,
+ "eval_accuracy": 0.8058,
+ "eval_loss": 0.578526904296875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 5984,
+ "train_examples": 45000,
+ "train_loss": 0.3893488602956136
+ },
+ {
+ "epoch": 18,
+ "eval_accuracy": 0.8182,
+ "eval_loss": 0.56311103515625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 6336,
+ "train_examples": 45000,
+ "train_loss": 0.38157559702661303
+ },
+ {
+ "epoch": 19,
+ "eval_accuracy": 0.8238,
+ "eval_loss": 0.5585554077148438,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 6688,
+ "train_examples": 45000,
+ "train_loss": 0.3700638519499037
+ },
+ {
+ "epoch": 20,
+ "eval_accuracy": 0.8182,
+ "eval_loss": 0.6097387573242188,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 7040,
+ "train_examples": 45000,
+ "train_loss": 0.3558312208970388
+ },
+ {
+ "epoch": 21,
+ "eval_accuracy": 0.8246,
+ "eval_loss": 0.5422500793457031,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 7392,
+ "train_examples": 45000,
+ "train_loss": 0.3505221447202894
+ },
+ {
+ "epoch": 22,
+ "eval_accuracy": 0.8226,
+ "eval_loss": 0.5890034423828125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 7744,
+ "train_examples": 45000,
+ "train_loss": 0.3476858967781067
+ },
+ {
+ "epoch": 23,
+ "eval_accuracy": 0.8584,
+ "eval_loss": 0.45730882568359377,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 8096,
+ "train_examples": 45000,
+ "train_loss": 0.3428149689250522
+ },
+ {
+ "epoch": 24,
+ "eval_accuracy": 0.8518,
+ "eval_loss": 0.48339465942382814,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 8448,
+ "train_examples": 45000,
+ "train_loss": 0.3287880612267388
+ },
+ {
+ "epoch": 25,
+ "eval_accuracy": 0.8414,
+ "eval_loss": 0.4980908935546875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 8800,
+ "train_examples": 45000,
+ "train_loss": 0.32596546165678236
+ },
+ {
+ "epoch": 26,
+ "eval_accuracy": 0.8368,
+ "eval_loss": 0.5045202209472657,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 9152,
+ "train_examples": 45000,
+ "train_loss": 0.31808142843776277
+ },
+ {
+ "epoch": 27,
+ "eval_accuracy": 0.8258,
+ "eval_loss": 0.5857073852539062,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 9504,
+ "train_examples": 45000,
+ "train_loss": 0.31575259125762517
+ },
+ {
+ "epoch": 28,
+ "eval_accuracy": 0.8596,
+ "eval_loss": 0.4352029235839844,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 9856,
+ "train_examples": 45000,
+ "train_loss": 0.30214522661368054
+ },
+ {
+ "epoch": 29,
+ "eval_accuracy": 0.8388,
+ "eval_loss": 0.5140209655761718,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 10208,
+ "train_examples": 45000,
+ "train_loss": 0.3071780029588275
+ },
+ {
+ "epoch": 30,
+ "eval_accuracy": 0.844,
+ "eval_loss": 0.5278230224609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 10560,
+ "train_examples": 45000,
+ "train_loss": 0.30251549474928113
+ },
+ {
+ "epoch": 31,
+ "eval_accuracy": 0.8484,
+ "eval_loss": 0.4901044494628906,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 10912,
+ "train_examples": 45000,
+ "train_loss": 0.2975416218333774
+ },
+ {
+ "epoch": 32,
+ "eval_accuracy": 0.8412,
+ "eval_loss": 0.49973211669921874,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 11264,
+ "train_examples": 45000,
+ "train_loss": 0.28743110587861803
+ },
+ {
+ "epoch": 33,
+ "eval_accuracy": 0.859,
+ "eval_loss": 0.45563653564453127,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 11616,
+ "train_examples": 45000,
+ "train_loss": 0.2848365594731437
+ },
+ {
+ "epoch": 34,
+ "eval_accuracy": 0.8556,
+ "eval_loss": 0.4716713928222656,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 11968,
+ "train_examples": 45000,
+ "train_loss": 0.2897914389504327
+ },
+ {
+ "epoch": 35,
+ "eval_accuracy": 0.8738,
+ "eval_loss": 0.4081214294433594,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 12320,
+ "train_examples": 45000,
+ "train_loss": 0.2804644786781735
+ },
+ {
+ "epoch": 36,
+ "eval_accuracy": 0.8702,
+ "eval_loss": 0.4287448791503906,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 12672,
+ "train_examples": 45000,
+ "train_loss": 0.27175004272990755
+ },
+ {
+ "epoch": 37,
+ "eval_accuracy": 0.8618,
+ "eval_loss": 0.4426470031738281,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 13024,
+ "train_examples": 45000,
+ "train_loss": 0.27865564083125854
+ },
+ {
+ "epoch": 38,
+ "eval_accuracy": 0.8396,
+ "eval_loss": 0.5234903503417969,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 13376,
+ "train_examples": 45000,
+ "train_loss": 0.26684371184772915
+ },
+ {
+ "epoch": 39,
+ "eval_accuracy": 0.855,
+ "eval_loss": 0.48744876708984375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 13728,
+ "train_examples": 45000,
+ "train_loss": 0.2674422241899702
+ },
+ {
+ "epoch": 40,
+ "eval_accuracy": 0.865,
+ "eval_loss": 0.4351408203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 14080,
+ "train_examples": 45000,
+ "train_loss": 0.2645341257254283
+ },
+ {
+ "epoch": 41,
+ "eval_accuracy": 0.8506,
+ "eval_loss": 0.4751714965820312,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 14432,
+ "train_examples": 45000,
+ "train_loss": 0.2622300498406092
+ },
+ {
+ "epoch": 42,
+ "eval_accuracy": 0.8486,
+ "eval_loss": 0.5085056274414063,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 14784,
+ "train_examples": 45000,
+ "train_loss": 0.25661233155727387
+ },
+ {
+ "epoch": 43,
+ "eval_accuracy": 0.8576,
+ "eval_loss": 0.4584455322265625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 15136,
+ "train_examples": 45000,
+ "train_loss": 0.256074648004108
+ },
+ {
+ "epoch": 44,
+ "eval_accuracy": 0.805,
+ "eval_loss": 0.6773380615234375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 15488,
+ "train_examples": 45000,
+ "train_loss": 0.25414007183975645
+ },
+ {
+ "epoch": 45,
+ "eval_accuracy": 0.867,
+ "eval_loss": 0.458577587890625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 15840,
+ "train_examples": 45000,
+ "train_loss": 0.24826880225075615
+ },
+ {
+ "epoch": 46,
+ "eval_accuracy": 0.8544,
+ "eval_loss": 0.47776199951171877,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 16192,
+ "train_examples": 45000,
+ "train_loss": 0.2539279241243998
+ },
+ {
+ "epoch": 47,
+ "eval_accuracy": 0.864,
+ "eval_loss": 0.44519176635742186,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 16544,
+ "train_examples": 45000,
+ "train_loss": 0.2478049456278483
+ },
+ {
+ "epoch": 48,
+ "eval_accuracy": 0.8556,
+ "eval_loss": 0.466750341796875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 16896,
+ "train_examples": 45000,
+ "train_loss": 0.24715948182211983
+ },
+ {
+ "epoch": 49,
+ "eval_accuracy": 0.8706,
+ "eval_loss": 0.45115664672851563,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 17248,
+ "train_examples": 45000,
+ "train_loss": 0.24359825609525046
+ },
+ {
+ "epoch": 50,
+ "eval_accuracy": 0.8546,
+ "eval_loss": 0.4703095520019531,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 17600,
+ "train_examples": 45000,
+ "train_loss": 0.24361659493313895
+ },
+ {
+ "epoch": 51,
+ "eval_accuracy": 0.8532,
+ "eval_loss": 0.5028222534179687,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 17952,
+ "train_examples": 45000,
+ "train_loss": 0.24154270447625054
+ },
+ {
+ "epoch": 52,
+ "eval_accuracy": 0.8768,
+ "eval_loss": 0.4012300231933594,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 18304,
+ "train_examples": 45000,
+ "train_loss": 0.23156300439834596
+ },
+ {
+ "epoch": 53,
+ "eval_accuracy": 0.8702,
+ "eval_loss": 0.43121511840820315,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 18656,
+ "train_examples": 45000,
+ "train_loss": 0.23431335463523864
+ },
+ {
+ "epoch": 54,
+ "eval_accuracy": 0.8506,
+ "eval_loss": 0.47547684326171874,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 19008,
+ "train_examples": 45000,
+ "train_loss": 0.23226308568848503
+ },
+ {
+ "epoch": 55,
+ "eval_accuracy": 0.8574,
+ "eval_loss": 0.47655361328125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 19360,
+ "train_examples": 45000,
+ "train_loss": 0.23324218198458355
+ },
+ {
+ "epoch": 56,
+ "eval_accuracy": 0.8666,
+ "eval_loss": 0.43212182006835936,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 19712,
+ "train_examples": 45000,
+ "train_loss": 0.23033563688066272
+ },
+ {
+ "epoch": 57,
+ "eval_accuracy": 0.8346,
+ "eval_loss": 0.5559804565429688,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 20064,
+ "train_examples": 45000,
+ "train_loss": 0.23134963205655415
+ },
+ {
+ "epoch": 58,
+ "eval_accuracy": 0.8638,
+ "eval_loss": 0.4514068420410156,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 20416,
+ "train_examples": 45000,
+ "train_loss": 0.2259135717921787
+ },
+ {
+ "epoch": 59,
+ "eval_accuracy": 0.8464,
+ "eval_loss": 0.4881473083496094,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 20768,
+ "train_examples": 45000,
+ "train_loss": 0.2296271209306187
+ },
+ {
+ "epoch": 60,
+ "eval_accuracy": 0.862,
+ "eval_loss": 0.48375531005859373,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 21120,
+ "train_examples": 45000,
+ "train_loss": 0.22572548664675818
+ },
+ {
+ "epoch": 61,
+ "eval_accuracy": 0.8684,
+ "eval_loss": 0.43897296142578124,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 21472,
+ "train_examples": 45000,
+ "train_loss": 0.22529616802533467
+ },
+ {
+ "epoch": 62,
+ "eval_accuracy": 0.8742,
+ "eval_loss": 0.39048966674804686,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 21824,
+ "train_examples": 45000,
+ "train_loss": 0.2210662999153137
+ },
+ {
+ "epoch": 63,
+ "eval_accuracy": 0.8618,
+ "eval_loss": 0.4357768005371094,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 22176,
+ "train_examples": 45000,
+ "train_loss": 0.2175708965169059
+ },
+ {
+ "epoch": 64,
+ "eval_accuracy": 0.8634,
+ "eval_loss": 0.46370264282226564,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 22528,
+ "train_examples": 45000,
+ "train_loss": 0.21798163765536413
+ },
+ {
+ "epoch": 65,
+ "eval_accuracy": 0.8574,
+ "eval_loss": 0.47452117309570313,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 22880,
+ "train_examples": 45000,
+ "train_loss": 0.22943376750946046
+ },
+ {
+ "epoch": 66,
+ "eval_accuracy": 0.8706,
+ "eval_loss": 0.42474490966796874,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 23232,
+ "train_examples": 45000,
+ "train_loss": 0.22070531824032466
+ },
+ {
+ "epoch": 67,
+ "eval_accuracy": 0.8738,
+ "eval_loss": 0.43662334594726565,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 23584,
+ "train_examples": 45000,
+ "train_loss": 0.21509841103288863
+ },
+ {
+ "epoch": 68,
+ "eval_accuracy": 0.8518,
+ "eval_loss": 0.47741646728515624,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 23936,
+ "train_examples": 45000,
+ "train_loss": 0.21424531423250834
+ },
+ {
+ "epoch": 69,
+ "eval_accuracy": 0.8728,
+ "eval_loss": 0.445956689453125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 24288,
+ "train_examples": 45000,
+ "train_loss": 0.20924902565744188
+ },
+ {
+ "epoch": 70,
+ "eval_accuracy": 0.8602,
+ "eval_loss": 0.4686448669433594,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 24640,
+ "train_examples": 45000,
+ "train_loss": 0.21512837672498492
+ },
+ {
+ "epoch": 71,
+ "eval_accuracy": 0.8564,
+ "eval_loss": 0.511124951171875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 24992,
+ "train_examples": 45000,
+ "train_loss": 0.2124501904964447
+ },
+ {
+ "epoch": 72,
+ "eval_accuracy": 0.8566,
+ "eval_loss": 0.4767055786132812,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 25344,
+ "train_examples": 45000,
+ "train_loss": 0.2073291993273629
+ },
+ {
+ "epoch": 73,
+ "eval_accuracy": 0.8534,
+ "eval_loss": 0.5192368408203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 25696,
+ "train_examples": 45000,
+ "train_loss": 0.21053057844903733
+ },
+ {
+ "epoch": 74,
+ "eval_accuracy": 0.8766,
+ "eval_loss": 0.38260752563476563,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 26048,
+ "train_examples": 45000,
+ "train_loss": 0.21595640751520792
+ },
+ {
+ "epoch": 75,
+ "eval_accuracy": 0.8756,
+ "eval_loss": 0.4221235717773438,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 26400,
+ "train_examples": 45000,
+ "train_loss": 0.21067103199164072
+ },
+ {
+ "epoch": 76,
+ "eval_accuracy": 0.876,
+ "eval_loss": 0.42229520263671877,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 26752,
+ "train_examples": 45000,
+ "train_loss": 0.20760168025758532
+ },
+ {
+ "epoch": 77,
+ "eval_accuracy": 0.8816,
+ "eval_loss": 0.41327566528320314,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 27104,
+ "train_examples": 45000,
+ "train_loss": 0.20539346924357943
+ },
+ {
+ "epoch": 78,
+ "eval_accuracy": 0.855,
+ "eval_loss": 0.4727892761230469,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 27456,
+ "train_examples": 45000,
+ "train_loss": 0.20733415001233418
+ },
+ {
+ "epoch": 79,
+ "eval_accuracy": 0.8538,
+ "eval_loss": 0.5247862976074219,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 27808,
+ "train_examples": 45000,
+ "train_loss": 0.20816828174591065
+ },
+ {
+ "epoch": 80,
+ "eval_accuracy": 0.8562,
+ "eval_loss": 0.5231691284179687,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 28160,
+ "train_examples": 45000,
+ "train_loss": 0.20593991249402363
+ },
+ {
+ "epoch": 81,
+ "eval_accuracy": 0.8644,
+ "eval_loss": 0.434323828125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 28512,
+ "train_examples": 45000,
+ "train_loss": 0.20798121780554454
+ },
+ {
+ "epoch": 82,
+ "eval_accuracy": 0.8718,
+ "eval_loss": 0.43466107177734375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 28864,
+ "train_examples": 45000,
+ "train_loss": 0.2018360275056627
+ },
+ {
+ "epoch": 83,
+ "eval_accuracy": 0.873,
+ "eval_loss": 0.4336739562988281,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 29216,
+ "train_examples": 45000,
+ "train_loss": 0.20214488338894315
+ },
+ {
+ "epoch": 84,
+ "eval_accuracy": 0.8738,
+ "eval_loss": 0.4181335754394531,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 29568,
+ "train_examples": 45000,
+ "train_loss": 0.20384543903138902
+ },
+ {
+ "epoch": 85,
+ "eval_accuracy": 0.8588,
+ "eval_loss": 0.5384344909667969,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 29920,
+ "train_examples": 45000,
+ "train_loss": 0.20373935463693407
+ },
+ {
+ "epoch": 86,
+ "eval_accuracy": 0.8714,
+ "eval_loss": 0.4508935974121094,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 30272,
+ "train_examples": 45000,
+ "train_loss": 0.20023504439459908
+ },
+ {
+ "epoch": 87,
+ "eval_accuracy": 0.8722,
+ "eval_loss": 0.4011263122558594,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 30624,
+ "train_examples": 45000,
+ "train_loss": 0.20357054704560174
+ },
+ {
+ "epoch": 88,
+ "eval_accuracy": 0.8724,
+ "eval_loss": 0.42971153564453124,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 30976,
+ "train_examples": 45000,
+ "train_loss": 0.2017909148454666
+ },
+ {
+ "epoch": 89,
+ "eval_accuracy": 0.8508,
+ "eval_loss": 0.5215956726074219,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 31328,
+ "train_examples": 45000,
+ "train_loss": 0.19414195964336395
+ },
+ {
+ "epoch": 90,
+ "eval_accuracy": 0.8584,
+ "eval_loss": 0.5140850219726563,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 31680,
+ "train_examples": 45000,
+ "train_loss": 0.19675261680814954
+ },
+ {
+ "epoch": 91,
+ "eval_accuracy": 0.8542,
+ "eval_loss": 0.5065567810058593,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 32032,
+ "train_examples": 45000,
+ "train_loss": 0.20285941220124562
+ },
+ {
+ "epoch": 92,
+ "eval_accuracy": 0.8632,
+ "eval_loss": 0.4859201049804687,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 32384,
+ "train_examples": 45000,
+ "train_loss": 0.19762985503408645
+ },
+ {
+ "epoch": 93,
+ "eval_accuracy": 0.8656,
+ "eval_loss": 0.4395970275878906,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 32736,
+ "train_examples": 45000,
+ "train_loss": 0.18883612869050767
+ },
+ {
+ "epoch": 94,
+ "eval_accuracy": 0.8754,
+ "eval_loss": 0.4072813049316406,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 33088,
+ "train_examples": 45000,
+ "train_loss": 0.20028164596822526
+ },
+ {
+ "epoch": 95,
+ "eval_accuracy": 0.876,
+ "eval_loss": 0.4278625915527344,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 33440,
+ "train_examples": 45000,
+ "train_loss": 0.18935567673047385
+ },
+ {
+ "epoch": 96,
+ "eval_accuracy": 0.8754,
+ "eval_loss": 0.448179248046875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 33792,
+ "train_examples": 45000,
+ "train_loss": 0.1924068394528495
+ },
+ {
+ "epoch": 97,
+ "eval_accuracy": 0.859,
+ "eval_loss": 0.4913785766601563,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 34144,
+ "train_examples": 45000,
+ "train_loss": 0.19636243831846448
+ },
+ {
+ "epoch": 98,
+ "eval_accuracy": 0.8798,
+ "eval_loss": 0.39225457153320314,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 34496,
+ "train_examples": 45000,
+ "train_loss": 0.19402525822851394
+ },
+ {
+ "epoch": 99,
+ "eval_accuracy": 0.8718,
+ "eval_loss": 0.4399384094238281,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 34848,
+ "train_examples": 45000,
+ "train_loss": 0.19180790162351397
+ },
+ {
+ "epoch": 100,
+ "eval_accuracy": 0.8756,
+ "eval_loss": 0.44355946044921873,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 35200,
+ "train_examples": 45000,
+ "train_loss": 0.19071120822164747
+ },
+ {
+ "epoch": 101,
+ "eval_accuracy": 0.9076,
+ "eval_loss": 0.306420263671875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 35552,
+ "train_examples": 45000,
+ "train_loss": 0.12239563805394703
+ },
+ {
+ "epoch": 102,
+ "eval_accuracy": 0.9104,
+ "eval_loss": 0.30657166137695313,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 35904,
+ "train_examples": 45000,
+ "train_loss": 0.09218944064643649
+ },
+ {
+ "epoch": 103,
+ "eval_accuracy": 0.9124,
+ "eval_loss": 0.298024169921875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 36256,
+ "train_examples": 45000,
+ "train_loss": 0.08602881506813898
+ },
+ {
+ "epoch": 104,
+ "eval_accuracy": 0.91,
+ "eval_loss": 0.30647501831054685,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 36608,
+ "train_examples": 45000,
+ "train_loss": 0.07542413108613756
+ },
+ {
+ "epoch": 105,
+ "eval_accuracy": 0.9144,
+ "eval_loss": 0.3046997314453125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 36960,
+ "train_examples": 45000,
+ "train_loss": 0.07289705408944024
+ },
+ {
+ "epoch": 106,
+ "eval_accuracy": 0.9158,
+ "eval_loss": 0.30186732177734377,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 37312,
+ "train_examples": 45000,
+ "train_loss": 0.06816188242700365
+ },
+ {
+ "epoch": 107,
+ "eval_accuracy": 0.9152,
+ "eval_loss": 0.3045905822753906,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 37664,
+ "train_examples": 45000,
+ "train_loss": 0.06303669462733799
+ },
+ {
+ "epoch": 108,
+ "eval_accuracy": 0.9138,
+ "eval_loss": 0.31312174072265625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 38016,
+ "train_examples": 45000,
+ "train_loss": 0.06031152280039258
+ },
+ {
+ "epoch": 109,
+ "eval_accuracy": 0.913,
+ "eval_loss": 0.3162181823730469,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 38368,
+ "train_examples": 45000,
+ "train_loss": 0.05866686734623379
+ },
+ {
+ "epoch": 110,
+ "eval_accuracy": 0.916,
+ "eval_loss": 0.3138496337890625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 38720,
+ "train_examples": 45000,
+ "train_loss": 0.0575275326596366
+ },
+ {
+ "epoch": 111,
+ "eval_accuracy": 0.9168,
+ "eval_loss": 0.31661144409179687,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 39072,
+ "train_examples": 45000,
+ "train_loss": 0.05273066160678864
+ },
+ {
+ "epoch": 112,
+ "eval_accuracy": 0.9194,
+ "eval_loss": 0.3204001953125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 39424,
+ "train_examples": 45000,
+ "train_loss": 0.05101389784415563
+ },
+ {
+ "epoch": 113,
+ "eval_accuracy": 0.9174,
+ "eval_loss": 0.33216910400390626,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 39776,
+ "train_examples": 45000,
+ "train_loss": 0.04935557192696465
+ },
+ {
+ "epoch": 114,
+ "eval_accuracy": 0.9158,
+ "eval_loss": 0.33527969970703125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 40128,
+ "train_examples": 45000,
+ "train_loss": 0.04651711027887132
+ },
+ {
+ "epoch": 115,
+ "eval_accuracy": 0.9148,
+ "eval_loss": 0.3381342224121094,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 40480,
+ "train_examples": 45000,
+ "train_loss": 0.046925569189919365
+ },
+ {
+ "epoch": 116,
+ "eval_accuracy": 0.9182,
+ "eval_loss": 0.33178223266601564,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 40832,
+ "train_examples": 45000,
+ "train_loss": 0.042672778570652006
+ },
+ {
+ "epoch": 117,
+ "eval_accuracy": 0.9158,
+ "eval_loss": 0.34461751098632815,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 41184,
+ "train_examples": 45000,
+ "train_loss": 0.044006208296616875
+ },
+ {
+ "epoch": 118,
+ "eval_accuracy": 0.917,
+ "eval_loss": 0.3422942260742188,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 41536,
+ "train_examples": 45000,
+ "train_loss": 0.04278108269870281
+ },
+ {
+ "epoch": 119,
+ "eval_accuracy": 0.9158,
+ "eval_loss": 0.34879398803710937,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 41888,
+ "train_examples": 45000,
+ "train_loss": 0.039726313829421996
+ },
+ {
+ "epoch": 120,
+ "eval_accuracy": 0.9192,
+ "eval_loss": 0.3475035583496094,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 42240,
+ "train_examples": 45000,
+ "train_loss": 0.0380223665383127
+ },
+ {
+ "epoch": 121,
+ "eval_accuracy": 0.9174,
+ "eval_loss": 0.35022559814453125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 42592,
+ "train_examples": 45000,
+ "train_loss": 0.037888774723476834
+ },
+ {
+ "epoch": 122,
+ "eval_accuracy": 0.9158,
+ "eval_loss": 0.34550585327148436,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 42944,
+ "train_examples": 45000,
+ "train_loss": 0.03741276666960783
+ },
+ {
+ "epoch": 123,
+ "eval_accuracy": 0.9174,
+ "eval_loss": 0.3567695190429688,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 43296,
+ "train_examples": 45000,
+ "train_loss": 0.03865679248174032
+ },
+ {
+ "epoch": 124,
+ "eval_accuracy": 0.9176,
+ "eval_loss": 0.35861453857421877,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 43648,
+ "train_examples": 45000,
+ "train_loss": 0.035508182628618344
+ },
+ {
+ "epoch": 125,
+ "eval_accuracy": 0.9162,
+ "eval_loss": 0.35652288208007815,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 44000,
+ "train_examples": 45000,
+ "train_loss": 0.036840546764598955
+ },
+ {
+ "epoch": 126,
+ "eval_accuracy": 0.918,
+ "eval_loss": 0.35606851806640627,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 44352,
+ "train_examples": 45000,
+ "train_loss": 0.03339896608259943
+ },
+ {
+ "epoch": 127,
+ "eval_accuracy": 0.9166,
+ "eval_loss": 0.36003599243164064,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 44704,
+ "train_examples": 45000,
+ "train_loss": 0.034402696437305876
+ },
+ {
+ "epoch": 128,
+ "eval_accuracy": 0.9166,
+ "eval_loss": 0.36662847900390627,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 45056,
+ "train_examples": 45000,
+ "train_loss": 0.03277938154472245
+ },
+ {
+ "epoch": 129,
+ "eval_accuracy": 0.9194,
+ "eval_loss": 0.36271561279296877,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 45408,
+ "train_examples": 45000,
+ "train_loss": 0.0318130090382364
+ },
+ {
+ "epoch": 130,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.36240728149414064,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 45760,
+ "train_examples": 45000,
+ "train_loss": 0.0301810850335492
+ },
+ {
+ "epoch": 131,
+ "eval_accuracy": 0.9156,
+ "eval_loss": 0.36919815063476563,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 46112,
+ "train_examples": 45000,
+ "train_loss": 0.03182596413758066
+ },
+ {
+ "epoch": 132,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3730984680175781,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 46464,
+ "train_examples": 45000,
+ "train_loss": 0.03062839696539773
+ },
+ {
+ "epoch": 133,
+ "eval_accuracy": 0.9164,
+ "eval_loss": 0.37771748046875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 46816,
+ "train_examples": 45000,
+ "train_loss": 0.028816458198097018
+ },
+ {
+ "epoch": 134,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.3720545104980469,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 47168,
+ "train_examples": 45000,
+ "train_loss": 0.029397612818082175
+ },
+ {
+ "epoch": 135,
+ "eval_accuracy": 0.9176,
+ "eval_loss": 0.3824631408691406,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 47520,
+ "train_examples": 45000,
+ "train_loss": 0.028310401866502233
+ },
+ {
+ "epoch": 136,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3975944519042969,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 47872,
+ "train_examples": 45000,
+ "train_loss": 0.027445427227682537
+ },
+ {
+ "epoch": 137,
+ "eval_accuracy": 0.9154,
+ "eval_loss": 0.388535009765625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 48224,
+ "train_examples": 45000,
+ "train_loss": 0.028050902290145556
+ },
+ {
+ "epoch": 138,
+ "eval_accuracy": 0.9144,
+ "eval_loss": 0.39537998046875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 48576,
+ "train_examples": 45000,
+ "train_loss": 0.027626405088106792
+ },
+ {
+ "epoch": 139,
+ "eval_accuracy": 0.9174,
+ "eval_loss": 0.3798079833984375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 48928,
+ "train_examples": 45000,
+ "train_loss": 0.02575553153918849
+ },
+ {
+ "epoch": 140,
+ "eval_accuracy": 0.9154,
+ "eval_loss": 0.3928522644042969,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 49280,
+ "train_examples": 45000,
+ "train_loss": 0.026475244792302448
+ },
+ {
+ "epoch": 141,
+ "eval_accuracy": 0.9176,
+ "eval_loss": 0.38615267944335935,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 49632,
+ "train_examples": 45000,
+ "train_loss": 0.026054457494947645
+ },
+ {
+ "epoch": 142,
+ "eval_accuracy": 0.9152,
+ "eval_loss": 0.3925371520996094,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 49984,
+ "train_examples": 45000,
+ "train_loss": 0.024948420998040174
+ },
+ {
+ "epoch": 143,
+ "eval_accuracy": 0.9142,
+ "eval_loss": 0.40529662475585937,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 50336,
+ "train_examples": 45000,
+ "train_loss": 0.025066455528471206
+ },
+ {
+ "epoch": 144,
+ "eval_accuracy": 0.9148,
+ "eval_loss": 0.3978242797851563,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 50688,
+ "train_examples": 45000,
+ "train_loss": 0.023501766060127154
+ },
+ {
+ "epoch": 145,
+ "eval_accuracy": 0.9146,
+ "eval_loss": 0.3887966552734375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 51040,
+ "train_examples": 45000,
+ "train_loss": 0.02534153993262185
+ },
+ {
+ "epoch": 146,
+ "eval_accuracy": 0.9182,
+ "eval_loss": 0.3958949401855469,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 51392,
+ "train_examples": 45000,
+ "train_loss": 0.025001446268293592
+ },
+ {
+ "epoch": 147,
+ "eval_accuracy": 0.9152,
+ "eval_loss": 0.3987830993652344,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 51744,
+ "train_examples": 45000,
+ "train_loss": 0.024140962322553
+ },
+ {
+ "epoch": 148,
+ "eval_accuracy": 0.9128,
+ "eval_loss": 0.3882256103515625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 52096,
+ "train_examples": 45000,
+ "train_loss": 0.02524253683288892
+ },
+ {
+ "epoch": 149,
+ "eval_accuracy": 0.9156,
+ "eval_loss": 0.39553275756835937,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 52448,
+ "train_examples": 45000,
+ "train_loss": 0.024576971726285087
+ },
+ {
+ "epoch": 150,
+ "eval_accuracy": 0.9168,
+ "eval_loss": 0.389462451171875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 52800,
+ "train_examples": 45000,
+ "train_loss": 0.022645456993414297
+ },
+ {
+ "epoch": 151,
+ "eval_accuracy": 0.9182,
+ "eval_loss": 0.38770169067382815,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 53152,
+ "train_examples": 45000,
+ "train_loss": 0.01990334998567899
+ },
+ {
+ "epoch": 152,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.38250919189453125,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 53504,
+ "train_examples": 45000,
+ "train_loss": 0.018938107203278275
+ },
+ {
+ "epoch": 153,
+ "eval_accuracy": 0.9152,
+ "eval_loss": 0.38870892333984375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 53856,
+ "train_examples": 45000,
+ "train_loss": 0.019380051201913093
+ },
+ {
+ "epoch": 154,
+ "eval_accuracy": 0.9174,
+ "eval_loss": 0.38606708984375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 54208,
+ "train_examples": 45000,
+ "train_loss": 0.01811076421704557
+ },
+ {
+ "epoch": 155,
+ "eval_accuracy": 0.917,
+ "eval_loss": 0.38509130859375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 54560,
+ "train_examples": 45000,
+ "train_loss": 0.017462889037529628
+ },
+ {
+ "epoch": 156,
+ "eval_accuracy": 0.9178,
+ "eval_loss": 0.3827947204589844,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 54912,
+ "train_examples": 45000,
+ "train_loss": 0.017325110151370365
+ },
+ {
+ "epoch": 157,
+ "eval_accuracy": 0.9174,
+ "eval_loss": 0.3864782897949219,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 55264,
+ "train_examples": 45000,
+ "train_loss": 0.017640994083881378
+ },
+ {
+ "epoch": 158,
+ "eval_accuracy": 0.9168,
+ "eval_loss": 0.3881909729003906,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 55616,
+ "train_examples": 45000,
+ "train_loss": 0.017731406083785824
+ },
+ {
+ "epoch": 159,
+ "eval_accuracy": 0.9178,
+ "eval_loss": 0.38386309204101565,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 55968,
+ "train_examples": 45000,
+ "train_loss": 0.01699283284942309
+ },
+ {
+ "epoch": 160,
+ "eval_accuracy": 0.9176,
+ "eval_loss": 0.38543032836914065,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 56320,
+ "train_examples": 45000,
+ "train_loss": 0.017252897179871798
+ },
+ {
+ "epoch": 161,
+ "eval_accuracy": 0.917,
+ "eval_loss": 0.3836103637695312,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 56672,
+ "train_examples": 45000,
+ "train_loss": 0.016278477271108163
+ },
+ {
+ "epoch": 162,
+ "eval_accuracy": 0.9164,
+ "eval_loss": 0.388681640625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 57024,
+ "train_examples": 45000,
+ "train_loss": 0.01604987183109754
+ },
+ {
+ "epoch": 163,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.38605330200195315,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 57376,
+ "train_examples": 45000,
+ "train_loss": 0.01634608430729972
+ },
+ {
+ "epoch": 164,
+ "eval_accuracy": 0.918,
+ "eval_loss": 0.38527861328125,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 57728,
+ "train_examples": 45000,
+ "train_loss": 0.015447496734228398
+ },
+ {
+ "epoch": 165,
+ "eval_accuracy": 0.9162,
+ "eval_loss": 0.39043060913085936,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 58080,
+ "train_examples": 45000,
+ "train_loss": 0.016543969903389612
+ },
+ {
+ "epoch": 166,
+ "eval_accuracy": 0.9168,
+ "eval_loss": 0.3898049377441406,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 58432,
+ "train_examples": 45000,
+ "train_loss": 0.016445220247407754
+ },
+ {
+ "epoch": 167,
+ "eval_accuracy": 0.917,
+ "eval_loss": 0.3887971130371094,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 58784,
+ "train_examples": 45000,
+ "train_loss": 0.016121218882666693
+ },
+ {
+ "epoch": 168,
+ "eval_accuracy": 0.9172,
+ "eval_loss": 0.3925276428222656,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 59136,
+ "train_examples": 45000,
+ "train_loss": 0.015302956238720153
+ },
+ {
+ "epoch": 169,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3891949768066406,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 59488,
+ "train_examples": 45000,
+ "train_loss": 0.01579265182117621
+ },
+ {
+ "epoch": 170,
+ "eval_accuracy": 0.916,
+ "eval_loss": 0.3865777526855469,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 59840,
+ "train_examples": 45000,
+ "train_loss": 0.015928712855113877
+ },
+ {
+ "epoch": 171,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3891114501953125,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 60192,
+ "train_examples": 45000,
+ "train_loss": 0.01610312660369608
+ },
+ {
+ "epoch": 172,
+ "eval_accuracy": 0.9178,
+ "eval_loss": 0.3897466430664063,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 60544,
+ "train_examples": 45000,
+ "train_loss": 0.015435847022467189
+ },
+ {
+ "epoch": 173,
+ "eval_accuracy": 0.917,
+ "eval_loss": 0.39024139404296876,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 60896,
+ "train_examples": 45000,
+ "train_loss": 0.016074572068452836
+ },
+ {
+ "epoch": 174,
+ "eval_accuracy": 0.9178,
+ "eval_loss": 0.3874936279296875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 61248,
+ "train_examples": 45000,
+ "train_loss": 0.01542677370963825
+ },
+ {
+ "epoch": 175,
+ "eval_accuracy": 0.9168,
+ "eval_loss": 0.3924413757324219,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 61600,
+ "train_examples": 45000,
+ "train_loss": 0.014539433761768871
+ },
+ {
+ "epoch": 176,
+ "eval_accuracy": 0.9176,
+ "eval_loss": 0.38622310791015624,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 61952,
+ "train_examples": 45000,
+ "train_loss": 0.01533671252793736
+ },
+ {
+ "epoch": 177,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.39207570190429686,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 62304,
+ "train_examples": 45000,
+ "train_loss": 0.01441938015388118
+ },
+ {
+ "epoch": 178,
+ "eval_accuracy": 0.9172,
+ "eval_loss": 0.394850341796875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 62656,
+ "train_examples": 45000,
+ "train_loss": 0.015913767455352678
+ },
+ {
+ "epoch": 179,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3906371459960937,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 63008,
+ "train_examples": 45000,
+ "train_loss": 0.015869515031576156
+ },
+ {
+ "epoch": 180,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3919496215820312,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 63360,
+ "train_examples": 45000,
+ "train_loss": 0.014408293558988306
+ },
+ {
+ "epoch": 181,
+ "eval_accuracy": 0.9196,
+ "eval_loss": 0.3897723571777344,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 63712,
+ "train_examples": 45000,
+ "train_loss": 0.01504027492403984
+ },
+ {
+ "epoch": 182,
+ "eval_accuracy": 0.9188,
+ "eval_loss": 0.3978993347167969,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 64064,
+ "train_examples": 45000,
+ "train_loss": 0.015615745646092628
+ },
+ {
+ "epoch": 183,
+ "eval_accuracy": 0.919,
+ "eval_loss": 0.39478623046875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 64416,
+ "train_examples": 45000,
+ "train_loss": 0.015364365304840935
+ },
+ {
+ "epoch": 184,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.39719932250976564,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 64768,
+ "train_examples": 45000,
+ "train_loss": 0.014254255270957947
+ },
+ {
+ "epoch": 185,
+ "eval_accuracy": 0.9188,
+ "eval_loss": 0.3978095458984375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 65120,
+ "train_examples": 45000,
+ "train_loss": 0.01493217945512798
+ },
+ {
+ "epoch": 186,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.39569725341796874,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 65472,
+ "train_examples": 45000,
+ "train_loss": 0.013478640712632073
+ },
+ {
+ "epoch": 187,
+ "eval_accuracy": 0.9166,
+ "eval_loss": 0.39535325317382813,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 65824,
+ "train_examples": 45000,
+ "train_loss": 0.013779820157090823
+ },
+ {
+ "epoch": 188,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3943298767089844,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 66176,
+ "train_examples": 45000,
+ "train_loss": 0.014935846232540078
+ },
+ {
+ "epoch": 189,
+ "eval_accuracy": 0.9178,
+ "eval_loss": 0.39513787841796877,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 66528,
+ "train_examples": 45000,
+ "train_loss": 0.014631074411628975
+ },
+ {
+ "epoch": 190,
+ "eval_accuracy": 0.917,
+ "eval_loss": 0.3951256713867187,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 66880,
+ "train_examples": 45000,
+ "train_loss": 0.01456148929099242
+ },
+ {
+ "epoch": 191,
+ "eval_accuracy": 0.9182,
+ "eval_loss": 0.3951748596191406,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 67232,
+ "train_examples": 45000,
+ "train_loss": 0.013960837727391885
+ },
+ {
+ "epoch": 192,
+ "eval_accuracy": 0.9178,
+ "eval_loss": 0.39580638427734377,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 67584,
+ "train_examples": 45000,
+ "train_loss": 0.014646381699293852
+ },
+ {
+ "epoch": 193,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3964273864746094,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 67936,
+ "train_examples": 45000,
+ "train_loss": 0.013860816383361816
+ },
+ {
+ "epoch": 194,
+ "eval_accuracy": 0.9186,
+ "eval_loss": 0.3953535705566406,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 68288,
+ "train_examples": 45000,
+ "train_loss": 0.013962972517146005
+ },
+ {
+ "epoch": 195,
+ "eval_accuracy": 0.9196,
+ "eval_loss": 0.3967518371582031,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 68640,
+ "train_examples": 45000,
+ "train_loss": 0.013807961952024037
+ },
+ {
+ "epoch": 196,
+ "eval_accuracy": 0.9182,
+ "eval_loss": 0.3972501220703125,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 68992,
+ "train_examples": 45000,
+ "train_loss": 0.014517988362577226
+ },
+ {
+ "epoch": 197,
+ "eval_accuracy": 0.9184,
+ "eval_loss": 0.3951761962890625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 69344,
+ "train_examples": 45000,
+ "train_loss": 0.01375881134768327
+ },
+ {
+ "epoch": 198,
+ "eval_accuracy": 0.9164,
+ "eval_loss": 0.39577669677734373,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 69696,
+ "train_examples": 45000,
+ "train_loss": 0.013727891698976358
+ },
+ {
+ "epoch": 199,
+ "eval_accuracy": 0.918,
+ "eval_loss": 0.3989869873046875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 70048,
+ "train_examples": 45000,
+ "train_loss": 0.014122135368155108
+ },
+ {
+ "epoch": 200,
+ "eval_accuracy": 0.9176,
+ "eval_loss": 0.39629163818359375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 70400,
+ "train_examples": 45000,
+ "train_loss": 0.014614876673618953
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 0,
+ "test_used_for_selection": false,
+ "validation_evaluations": 201
+ },
+ "final": {
+ "accuracy": 0.9176,
+ "epoch": 200,
+ "evaluation_split": "validation",
+ "finite": true,
+ "loss": 0.39629163818359375,
+ "step": 70400
+ },
+ "hardware": {
+ "cuda_device_name": "NVIDIA RTX A6000",
+ "cuda_visible_devices": "0",
+ "device": "cuda",
+ "device_total_memory_bytes": 50896568320,
+ "peak_memory_allocated_bytes": 2017119744,
+ "peak_memory_reserved_bytes": 2464153600,
+ "torch_version": "2.3.1+cu118"
+ },
+ "protocol_family": "oral_a_cifar_local_resnet_development",
+ "provenance": {
+ "git_commit": "f55872beaee658ff76ac15fd421b73017820c320",
+ "git_tracked_dirty": false
+ },
+ "schema_version": 1,
+ "split": {
+ "cifar_source_files": [
+ {
+ "bytes": 31035704,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_1",
+ "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
+ },
+ {
+ "bytes": 31035320,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_2",
+ "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
+ },
+ {
+ "bytes": 31035999,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_3",
+ "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
+ },
+ {
+ "bytes": 31035696,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_4",
+ "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
+ },
+ {
+ "bytes": 31035623,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_5",
+ "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
+ },
+ {
+ "bytes": 31035526,
+ "path": "/orion/u/oscarwan/scratch/sdil-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": 21.029086351394653,
+ "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": 1224.873178243637,
+ "train_wall_s": 1203.823739528656
+ },
+ "work": {
+ "apical_macs_per_example": 0,
+ "causal_scalar_observations": 0,
+ "components": {
+ "apical_projection_macs": 0,
+ "apical_regression_macs": 0,
+ "bp_reverse_macs_estimate": 729918720000000,
+ "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": 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; 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": 40551040,
+ "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": 9000000,
+ "total_forward_equivalent_examples": 9000000,
+ "total_macs_estimate": 1094878080000000
+ }
+}
generated by cgit v1.2.3 (git 2.25.1) at 2026-09-16 19:18:34 +0000