summaryrefslogtreecommitdiff
path: root/results/resnet_crossover/r2/resnet-r2-fa-d32-lr0p1.json
diff options
context:
space:
mode:
Diffstat (limited to 'results/resnet_crossover/r2/resnet-r2-fa-d32-lr0p1.json')
-rw-r--r--results/resnet_crossover/r2/resnet-r2-fa-d32-lr0p1.json2801
1 files changed, 2801 insertions, 0 deletions
diff --git a/results/resnet_crossover/r2/resnet-r2-fa-d32-lr0p1.json b/results/resnet_crossover/r2/resnet-r2-fa-d32-lr0p1.json
new file mode 100644
index 0000000..a2f3e08
--- /dev/null
+++ b/results/resnet_crossover/r2/resnet-r2-fa-d32-lr0p1.json
@@ -0,0 +1,2801 @@
+{
+ "architecture": {
+ "adaptive_apical_parameters": 0,
+ "adaptive_feedback_parameters": 0,
+ "base_width": 16,
+ "blocks_per_stage": 5,
+ "bn_eps": 1e-05,
+ "bn_momentum": 0.1,
+ "depth": 32,
+ "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+ "fixed_feedback_parameters": 461440,
+ "fixed_traffic_coefficients": 0,
+ "forward_parameters": 464154,
+ "hidden_shapes": [
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 16,
+ 32,
+ 32
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 32,
+ 16,
+ 16
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ]
+ ],
+ "normalization": "batchnorm",
+ "predictor_parameters": 0,
+ "residual_scale": 1.0,
+ "vectorizer_mode": null,
+ "vectorizer_parameters": 0
+ },
+ "args": {
+ "a_scale": 1.0,
+ "a_warmup_steps": 0,
+ "alignment_probe": 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": "/orion/u/oscarwan/scratch/sdil-data",
+ "depth": 32,
+ "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": "hfa",
+ "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-fa-d32-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": {
+ "early_third_mean": 0.03778243931010365,
+ "feedback_forward_cosine": [
+ 0.13522903621196747,
+ 0.2389027625322342,
+ 0.1656235307455063,
+ 0.25669336318969727,
+ 0.08168920129537582,
+ 0.24313712120056152,
+ 0.03679082542657852,
+ 0.16495703160762787,
+ 0.043550122529268265,
+ 0.20206722617149353,
+ 0.052305079996585846,
+ 0.1600673943758011,
+ 0.04140310361981392,
+ 0.1889246553182602,
+ 0.029380766674876213,
+ 0.2227826863527298,
+ 0.028732571750879288,
+ 0.18993735313415527,
+ 0.032700490206480026,
+ 0.19198478758335114,
+ 0.031028691679239273,
+ 0.1495836079120636,
+ 0.017455775290727615,
+ 0.18182963132858276,
+ 0.03303111344575882,
+ 0.18500036001205444,
+ 0.003500960301607847,
+ 0.19491636753082275,
+ 0.02336471900343895,
+ 0.15348230302333832,
+ 0.7153934240341187
+ ],
+ "feedback_forward_norm_ratio": [
+ 1.6532303094863892,
+ 1.3528220653533936,
+ 1.8407398462295532,
+ 1.573699712753296,
+ 2.2097935676574707,
+ 1.9533603191375732,
+ 2.9062178134918213,
+ 2.332139253616333,
+ 2.6777613162994385,
+ 2.2939107418060303,
+ 2.361605405807495,
+ 1.6632351875305176,
+ 2.157731771469116,
+ 1.681369423866272,
+ 2.3610785007476807,
+ 1.9888919591903687,
+ 2.5523250102996826,
+ 2.211498737335205,
+ 2.42690110206604,
+ 2.11393141746521,
+ 2.266143560409546,
+ 1.3281869888305664,
+ 2.0636491775512695,
+ 1.2860081195831299,
+ 1.958283543586731,
+ 1.3220630884170532,
+ 1.8889795541763306,
+ 1.3131144046783447,
+ 2.109546184539795,
+ 1.3858412504196167,
+ 0.19881561398506165
+ ],
+ "feedback_forward_relative_error": [
+ 1.8127440214157104,
+ 1.477748990058899,
+ 1.9438579082489014,
+ 1.6335893869400024,
+ 2.349926710128784,
+ 1.9661505222320557,
+ 3.0384628772735596,
+ 2.3810644149780273,
+ 2.817298650741577,
+ 2.3097572326660156,
+ 2.5159752368927,
+ 1.7983025312423706,
+ 2.340327739715576,
+ 1.7865326404571533,
+ 2.5369174480438232,
+ 2.0173025131225586,
+ 2.7143495082855225,
+ 2.247361660003662,
+ 2.5944416522979736,
+ 2.158013343811035,
+ 2.448423147201538,
+ 1.5384182929992676,
+ 2.277411460876465,
+ 1.4785629510879517,
+ 2.169217586517334,
+ 1.5028926134109497,
+ 2.134248733520508,
+ 1.4874053001403809,
+ 2.3133537769317627,
+ 1.5796048641204834,
+ 0.8689447045326233
+ ],
+ "innovation_negative_gradient_cosine": [
+ 0.00606947299093008,
+ 0.009478073567152023,
+ 0.024084236472845078,
+ 0.022062642499804497,
+ 0.04974904656410217,
+ 0.037176258862018585,
+ 0.07845072448253632,
+ 0.028204817324876785,
+ 0.09064127504825592,
+ 0.03190784528851509,
+ 0.09537523239850998,
+ 0.06018902733922005,
+ 0.18250532448291779,
+ 0.06845758110284805,
+ 0.22662648558616638,
+ 0.09742648154497147,
+ 0.2728409171104431,
+ 0.07810325175523758,
+ 0.29824674129486084,
+ 0.08288314938545227,
+ 0.3066348731517792,
+ 0.09523385763168335,
+ 0.4438329339027405,
+ 0.10535746812820435,
+ 0.4930099844932556,
+ 0.13920406997203827,
+ 0.5292818546295166,
+ 0.2641158998012543,
+ 0.6173524260520935,
+ 0.31898194551467896,
+ 0.7836034893989563
+ ],
+ "mean_feedback_forward_cosine": 0.1417885826930644,
+ "mean_feedback_forward_relative_error": 2.072213174835328,
+ "normalization_state": "training_batch_stats_without_running_update",
+ "raw_negative_gradient_cosine": [
+ 0.00606947299093008,
+ 0.009478073567152023,
+ 0.024084236472845078,
+ 0.022062642499804497,
+ 0.04974904656410217,
+ 0.037176258862018585,
+ 0.07845072448253632,
+ 0.028204817324876785,
+ 0.09064127504825592,
+ 0.03190784528851509,
+ 0.09537523239850998,
+ 0.06018902733922005,
+ 0.18250532448291779,
+ 0.06845758110284805,
+ 0.22662648558616638,
+ 0.09742648154497147,
+ 0.2728409171104431,
+ 0.07810325175523758,
+ 0.29824674129486084,
+ 0.08288314938545227,
+ 0.3066348731517792,
+ 0.09523385763168335,
+ 0.4438329339027405,
+ 0.10535746812820435,
+ 0.4930099844932556,
+ 0.13920406997203827,
+ 0.5292818546295166,
+ 0.2641158998012543,
+ 0.6173524260520935,
+ 0.31898194551467896,
+ 0.7836034893989563
+ ],
+ "teaching_negative_gradient_cosine": [
+ 0.00606947299093008,
+ 0.009478073567152023,
+ 0.024084236472845078,
+ 0.022062642499804497,
+ 0.04974904656410217,
+ 0.037176258862018585,
+ 0.07845072448253632,
+ 0.028204817324876785,
+ 0.09064127504825592,
+ 0.03190784528851509,
+ 0.09537523239850998,
+ 0.06018902733922005,
+ 0.18250532448291779,
+ 0.06845758110284805,
+ 0.22662648558616638,
+ 0.09742648154497147,
+ 0.2728409171104431,
+ 0.07810325175523758,
+ 0.29824674129486084,
+ 0.08288314938545227,
+ 0.3066348731517792,
+ 0.09523385763168335,
+ 0.4438329339027405,
+ 0.10535746812820435,
+ 0.4930099844932556,
+ 0.13920406997203827,
+ 0.5292818546295166,
+ 0.2641158998012543,
+ 0.6173524260520935,
+ 0.31898194551467896,
+ 0.7836034893989563
+ ],
+ "wall_s": 0.11364865303039551
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "eval_accuracy": 0.2956,
+ "eval_loss": 1.986936767578125,
+ "eval_split": "validation",
+ "lr": 0.02,
+ "output_lr": 0.02,
+ "step": 352,
+ "train_examples": 45000,
+ "train_loss": 2.195509643088447
+ },
+ {
+ "epoch": 2,
+ "eval_accuracy": 0.2702,
+ "eval_loss": 2.53878330078125,
+ "eval_split": "validation",
+ "lr": 0.04,
+ "output_lr": 0.04,
+ "step": 704,
+ "train_examples": 45000,
+ "train_loss": 2.3834243949042424
+ },
+ {
+ "epoch": 3,
+ "eval_accuracy": 0.2754,
+ "eval_loss": 3.973369384765625,
+ "eval_split": "validation",
+ "lr": 0.06000000000000001,
+ "output_lr": 0.06000000000000001,
+ "step": 1056,
+ "train_examples": 45000,
+ "train_loss": 3.8713371835496693
+ },
+ {
+ "epoch": 4,
+ "eval_accuracy": 0.272,
+ "eval_loss": 7.13771689453125,
+ "eval_split": "validation",
+ "lr": 0.08,
+ "output_lr": 0.08,
+ "step": 1408,
+ "train_examples": 45000,
+ "train_loss": 6.064967641448975
+ },
+ {
+ "epoch": 5,
+ "eval_accuracy": 0.2578,
+ "eval_loss": 9.420830078125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 1760,
+ "train_examples": 45000,
+ "train_loss": 7.330231070963542
+ },
+ {
+ "epoch": 6,
+ "eval_accuracy": 0.3024,
+ "eval_loss": 7.5071580078125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 2112,
+ "train_examples": 45000,
+ "train_loss": 7.459590669589573
+ },
+ {
+ "epoch": 7,
+ "eval_accuracy": 0.341,
+ "eval_loss": 9.0689892578125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 2464,
+ "train_examples": 45000,
+ "train_loss": 6.910371139611138
+ },
+ {
+ "epoch": 8,
+ "eval_accuracy": 0.2986,
+ "eval_loss": 7.185187890625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 2816,
+ "train_examples": 45000,
+ "train_loss": 7.848369245402019
+ },
+ {
+ "epoch": 9,
+ "eval_accuracy": 0.268,
+ "eval_loss": 7.8466748046875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 3168,
+ "train_examples": 45000,
+ "train_loss": 7.287076099565294
+ },
+ {
+ "epoch": 10,
+ "eval_accuracy": 0.3114,
+ "eval_loss": 6.8664982421875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 3520,
+ "train_examples": 45000,
+ "train_loss": 6.893469244893392
+ },
+ {
+ "epoch": 11,
+ "eval_accuracy": 0.2968,
+ "eval_loss": 6.61613349609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 3872,
+ "train_examples": 45000,
+ "train_loss": 7.038937747870551
+ },
+ {
+ "epoch": 12,
+ "eval_accuracy": 0.3484,
+ "eval_loss": 8.7687521484375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 4224,
+ "train_examples": 45000,
+ "train_loss": 6.735957527669271
+ },
+ {
+ "epoch": 13,
+ "eval_accuracy": 0.3388,
+ "eval_loss": 6.116796875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 4576,
+ "train_examples": 45000,
+ "train_loss": 6.188509490797255
+ },
+ {
+ "epoch": 14,
+ "eval_accuracy": 0.3352,
+ "eval_loss": 6.60065009765625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 4928,
+ "train_examples": 45000,
+ "train_loss": 6.898669404178196
+ },
+ {
+ "epoch": 15,
+ "eval_accuracy": 0.3442,
+ "eval_loss": 7.7424158203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 5280,
+ "train_examples": 45000,
+ "train_loss": 6.403976938290066
+ },
+ {
+ "epoch": 16,
+ "eval_accuracy": 0.3778,
+ "eval_loss": 6.02613154296875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 5632,
+ "train_examples": 45000,
+ "train_loss": 6.261937475925022
+ },
+ {
+ "epoch": 17,
+ "eval_accuracy": 0.3454,
+ "eval_loss": 6.16180947265625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 5984,
+ "train_examples": 45000,
+ "train_loss": 5.652168246968587
+ },
+ {
+ "epoch": 18,
+ "eval_accuracy": 0.2904,
+ "eval_loss": 12.1865435546875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 6336,
+ "train_examples": 45000,
+ "train_loss": 5.759846517774794
+ },
+ {
+ "epoch": 19,
+ "eval_accuracy": 0.3586,
+ "eval_loss": 6.29294453125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 6688,
+ "train_examples": 45000,
+ "train_loss": 5.43130024939643
+ },
+ {
+ "epoch": 20,
+ "eval_accuracy": 0.3414,
+ "eval_loss": 6.084059375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 7040,
+ "train_examples": 45000,
+ "train_loss": 5.368133390469021
+ },
+ {
+ "epoch": 21,
+ "eval_accuracy": 0.3324,
+ "eval_loss": 7.98075751953125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 7392,
+ "train_examples": 45000,
+ "train_loss": 5.172292945861816
+ },
+ {
+ "epoch": 22,
+ "eval_accuracy": 0.3692,
+ "eval_loss": 6.4959990234375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 7744,
+ "train_examples": 45000,
+ "train_loss": 5.412291467454699
+ },
+ {
+ "epoch": 23,
+ "eval_accuracy": 0.362,
+ "eval_loss": 6.58297880859375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 8096,
+ "train_examples": 45000,
+ "train_loss": 5.337951386176215
+ },
+ {
+ "epoch": 24,
+ "eval_accuracy": 0.3838,
+ "eval_loss": 5.59174921875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 8448,
+ "train_examples": 45000,
+ "train_loss": 5.547587069659763
+ },
+ {
+ "epoch": 25,
+ "eval_accuracy": 0.3284,
+ "eval_loss": 8.17561748046875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 8800,
+ "train_examples": 45000,
+ "train_loss": 5.564922823164198
+ },
+ {
+ "epoch": 26,
+ "eval_accuracy": 0.4088,
+ "eval_loss": 5.76070908203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 9152,
+ "train_examples": 45000,
+ "train_loss": 5.24930276345147
+ },
+ {
+ "epoch": 27,
+ "eval_accuracy": 0.3016,
+ "eval_loss": 7.5848359375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 9504,
+ "train_examples": 45000,
+ "train_loss": 5.235181749979655
+ },
+ {
+ "epoch": 28,
+ "eval_accuracy": 0.3864,
+ "eval_loss": 6.51845185546875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 9856,
+ "train_examples": 45000,
+ "train_loss": 5.172612402937148
+ },
+ {
+ "epoch": 29,
+ "eval_accuracy": 0.3368,
+ "eval_loss": 7.98949931640625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 10208,
+ "train_examples": 45000,
+ "train_loss": 5.318213782840305
+ },
+ {
+ "epoch": 30,
+ "eval_accuracy": 0.3612,
+ "eval_loss": 8.5869908203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 10560,
+ "train_examples": 45000,
+ "train_loss": 5.080642896185981
+ },
+ {
+ "epoch": 31,
+ "eval_accuracy": 0.3554,
+ "eval_loss": 6.6961919921875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 10912,
+ "train_examples": 45000,
+ "train_loss": 5.069903329298231
+ },
+ {
+ "epoch": 32,
+ "eval_accuracy": 0.3928,
+ "eval_loss": 6.05183935546875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 11264,
+ "train_examples": 45000,
+ "train_loss": 5.545271482086181
+ },
+ {
+ "epoch": 33,
+ "eval_accuracy": 0.3728,
+ "eval_loss": 7.6160564453125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 11616,
+ "train_examples": 45000,
+ "train_loss": 5.059899168311225
+ },
+ {
+ "epoch": 34,
+ "eval_accuracy": 0.4254,
+ "eval_loss": 4.392253515625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 11968,
+ "train_examples": 45000,
+ "train_loss": 4.735173885133531
+ },
+ {
+ "epoch": 35,
+ "eval_accuracy": 0.4154,
+ "eval_loss": 5.35446240234375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 12320,
+ "train_examples": 45000,
+ "train_loss": 4.707660108100043
+ },
+ {
+ "epoch": 36,
+ "eval_accuracy": 0.4556,
+ "eval_loss": 5.477988671875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 12672,
+ "train_examples": 45000,
+ "train_loss": 4.8679831588745115
+ },
+ {
+ "epoch": 37,
+ "eval_accuracy": 0.3956,
+ "eval_loss": 5.809548828125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 13024,
+ "train_examples": 45000,
+ "train_loss": 5.061632610660129
+ },
+ {
+ "epoch": 38,
+ "eval_accuracy": 0.38,
+ "eval_loss": 8.34639951171875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 13376,
+ "train_examples": 45000,
+ "train_loss": 5.17016610599094
+ },
+ {
+ "epoch": 39,
+ "eval_accuracy": 0.3596,
+ "eval_loss": 10.3157998046875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 13728,
+ "train_examples": 45000,
+ "train_loss": 5.13461835738288
+ },
+ {
+ "epoch": 40,
+ "eval_accuracy": 0.4494,
+ "eval_loss": 3.57537568359375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 14080,
+ "train_examples": 45000,
+ "train_loss": 4.814193962224325
+ },
+ {
+ "epoch": 41,
+ "eval_accuracy": 0.3972,
+ "eval_loss": 7.6974388671875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 14432,
+ "train_examples": 45000,
+ "train_loss": 4.812456588575575
+ },
+ {
+ "epoch": 42,
+ "eval_accuracy": 0.4116,
+ "eval_loss": 5.53400478515625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 14784,
+ "train_examples": 45000,
+ "train_loss": 4.879522632090251
+ },
+ {
+ "epoch": 43,
+ "eval_accuracy": 0.3586,
+ "eval_loss": 7.28328564453125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 15136,
+ "train_examples": 45000,
+ "train_loss": 4.666603200615777
+ },
+ {
+ "epoch": 44,
+ "eval_accuracy": 0.4092,
+ "eval_loss": 5.62242421875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 15488,
+ "train_examples": 45000,
+ "train_loss": 5.149397599199083
+ },
+ {
+ "epoch": 45,
+ "eval_accuracy": 0.4616,
+ "eval_loss": 4.175123388671875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 15840,
+ "train_examples": 45000,
+ "train_loss": 4.626690245098538
+ },
+ {
+ "epoch": 46,
+ "eval_accuracy": 0.4692,
+ "eval_loss": 5.34783779296875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 16192,
+ "train_examples": 45000,
+ "train_loss": 4.866097333908081
+ },
+ {
+ "epoch": 47,
+ "eval_accuracy": 0.4406,
+ "eval_loss": 5.89131806640625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 16544,
+ "train_examples": 45000,
+ "train_loss": 4.398048905860053
+ },
+ {
+ "epoch": 48,
+ "eval_accuracy": 0.4358,
+ "eval_loss": 5.7748138671875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 16896,
+ "train_examples": 45000,
+ "train_loss": 4.245615175713433
+ },
+ {
+ "epoch": 49,
+ "eval_accuracy": 0.5006,
+ "eval_loss": 4.4333083984375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 17248,
+ "train_examples": 45000,
+ "train_loss": 4.748879802788628
+ },
+ {
+ "epoch": 50,
+ "eval_accuracy": 0.4424,
+ "eval_loss": 4.04569853515625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 17600,
+ "train_examples": 45000,
+ "train_loss": 4.1075999011993405
+ },
+ {
+ "epoch": 51,
+ "eval_accuracy": 0.4298,
+ "eval_loss": 4.8061736328125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 17952,
+ "train_examples": 45000,
+ "train_loss": 4.082825281778971
+ },
+ {
+ "epoch": 52,
+ "eval_accuracy": 0.3572,
+ "eval_loss": 5.676712109375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 18304,
+ "train_examples": 45000,
+ "train_loss": 4.619352707672119
+ },
+ {
+ "epoch": 53,
+ "eval_accuracy": 0.423,
+ "eval_loss": 6.05435361328125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 18656,
+ "train_examples": 45000,
+ "train_loss": 3.9728969636705185
+ },
+ {
+ "epoch": 54,
+ "eval_accuracy": 0.4374,
+ "eval_loss": 4.6374767578125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 19008,
+ "train_examples": 45000,
+ "train_loss": 4.3532811665005156
+ },
+ {
+ "epoch": 55,
+ "eval_accuracy": 0.3908,
+ "eval_loss": 5.56046455078125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 19360,
+ "train_examples": 45000,
+ "train_loss": 3.9575316127353246
+ },
+ {
+ "epoch": 56,
+ "eval_accuracy": 0.4264,
+ "eval_loss": 4.6221392578125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 19712,
+ "train_examples": 45000,
+ "train_loss": 4.364873327128093
+ },
+ {
+ "epoch": 57,
+ "eval_accuracy": 0.4556,
+ "eval_loss": 3.86793564453125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 20064,
+ "train_examples": 45000,
+ "train_loss": 3.9743894003974067
+ },
+ {
+ "epoch": 58,
+ "eval_accuracy": 0.401,
+ "eval_loss": 5.83616787109375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 20416,
+ "train_examples": 45000,
+ "train_loss": 4.274858414755927
+ },
+ {
+ "epoch": 59,
+ "eval_accuracy": 0.4062,
+ "eval_loss": 5.9770345703125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 20768,
+ "train_examples": 45000,
+ "train_loss": 4.168988529247708
+ },
+ {
+ "epoch": 60,
+ "eval_accuracy": 0.4162,
+ "eval_loss": 5.7100212890625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 21120,
+ "train_examples": 45000,
+ "train_loss": 4.416859284888374
+ },
+ {
+ "epoch": 61,
+ "eval_accuracy": 0.4016,
+ "eval_loss": 4.9233658203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 21472,
+ "train_examples": 45000,
+ "train_loss": 4.03270875193278
+ },
+ {
+ "epoch": 62,
+ "eval_accuracy": 0.41,
+ "eval_loss": 7.848937890625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 21824,
+ "train_examples": 45000,
+ "train_loss": 4.23637970928616
+ },
+ {
+ "epoch": 63,
+ "eval_accuracy": 0.4404,
+ "eval_loss": 8.1851318359375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 22176,
+ "train_examples": 45000,
+ "train_loss": 4.32700122680664
+ },
+ {
+ "epoch": 64,
+ "eval_accuracy": 0.4304,
+ "eval_loss": 5.8247,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 22528,
+ "train_examples": 45000,
+ "train_loss": 4.107528818427192
+ },
+ {
+ "epoch": 65,
+ "eval_accuracy": 0.4472,
+ "eval_loss": 6.40511494140625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 22880,
+ "train_examples": 45000,
+ "train_loss": 4.552944485388862
+ },
+ {
+ "epoch": 66,
+ "eval_accuracy": 0.395,
+ "eval_loss": 6.11930693359375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 23232,
+ "train_examples": 45000,
+ "train_loss": 4.474366613896688
+ },
+ {
+ "epoch": 67,
+ "eval_accuracy": 0.4084,
+ "eval_loss": 6.2334013671875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 23584,
+ "train_examples": 45000,
+ "train_loss": 4.986977877638075
+ },
+ {
+ "epoch": 68,
+ "eval_accuracy": 0.4232,
+ "eval_loss": 5.4462650390625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 23936,
+ "train_examples": 45000,
+ "train_loss": 4.605135821109348
+ },
+ {
+ "epoch": 69,
+ "eval_accuracy": 0.3384,
+ "eval_loss": 6.6896943359375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 24288,
+ "train_examples": 45000,
+ "train_loss": 4.689142846934001
+ },
+ {
+ "epoch": 70,
+ "eval_accuracy": 0.4314,
+ "eval_loss": 7.1887201171875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 24640,
+ "train_examples": 45000,
+ "train_loss": 4.31898379787869
+ },
+ {
+ "epoch": 71,
+ "eval_accuracy": 0.4278,
+ "eval_loss": 7.44157431640625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 24992,
+ "train_examples": 45000,
+ "train_loss": 4.600417254977756
+ },
+ {
+ "epoch": 72,
+ "eval_accuracy": 0.409,
+ "eval_loss": 6.05228955078125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 25344,
+ "train_examples": 45000,
+ "train_loss": 4.883821003723145
+ },
+ {
+ "epoch": 73,
+ "eval_accuracy": 0.4002,
+ "eval_loss": 6.30699609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 25696,
+ "train_examples": 45000,
+ "train_loss": 4.3505779678344725
+ },
+ {
+ "epoch": 74,
+ "eval_accuracy": 0.4464,
+ "eval_loss": 6.32979833984375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 26048,
+ "train_examples": 45000,
+ "train_loss": 4.398072329033746
+ },
+ {
+ "epoch": 75,
+ "eval_accuracy": 0.4112,
+ "eval_loss": 6.6822724609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 26400,
+ "train_examples": 45000,
+ "train_loss": 4.89097486360338
+ },
+ {
+ "epoch": 76,
+ "eval_accuracy": 0.4596,
+ "eval_loss": 4.2010662109375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 26752,
+ "train_examples": 45000,
+ "train_loss": 4.577514958106147
+ },
+ {
+ "epoch": 77,
+ "eval_accuracy": 0.3782,
+ "eval_loss": 9.2190607421875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 27104,
+ "train_examples": 45000,
+ "train_loss": 4.218928383212619
+ },
+ {
+ "epoch": 78,
+ "eval_accuracy": 0.4284,
+ "eval_loss": 9.0243501953125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 27456,
+ "train_examples": 45000,
+ "train_loss": 4.1683951346503365
+ },
+ {
+ "epoch": 79,
+ "eval_accuracy": 0.3524,
+ "eval_loss": 8.21492509765625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 27808,
+ "train_examples": 45000,
+ "train_loss": 4.540081050364177
+ },
+ {
+ "epoch": 80,
+ "eval_accuracy": 0.406,
+ "eval_loss": 6.62808154296875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 28160,
+ "train_examples": 45000,
+ "train_loss": 4.4213002469380696
+ },
+ {
+ "epoch": 81,
+ "eval_accuracy": 0.4084,
+ "eval_loss": 7.6263900390625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 28512,
+ "train_examples": 45000,
+ "train_loss": 4.814154179806179
+ },
+ {
+ "epoch": 82,
+ "eval_accuracy": 0.4404,
+ "eval_loss": 4.99802724609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 28864,
+ "train_examples": 45000,
+ "train_loss": 4.313245312457615
+ },
+ {
+ "epoch": 83,
+ "eval_accuracy": 0.4326,
+ "eval_loss": 4.130868115234375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 29216,
+ "train_examples": 45000,
+ "train_loss": 4.179187333128187
+ },
+ {
+ "epoch": 84,
+ "eval_accuracy": 0.4598,
+ "eval_loss": 6.097021484375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 29568,
+ "train_examples": 45000,
+ "train_loss": 4.335098880301581
+ },
+ {
+ "epoch": 85,
+ "eval_accuracy": 0.4,
+ "eval_loss": 6.01697724609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 29920,
+ "train_examples": 45000,
+ "train_loss": 4.538031046464708
+ },
+ {
+ "epoch": 86,
+ "eval_accuracy": 0.4672,
+ "eval_loss": 5.2787099609375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 30272,
+ "train_examples": 45000,
+ "train_loss": 4.391405413013034
+ },
+ {
+ "epoch": 87,
+ "eval_accuracy": 0.328,
+ "eval_loss": 7.990158203125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 30624,
+ "train_examples": 45000,
+ "train_loss": 4.530714601389567
+ },
+ {
+ "epoch": 88,
+ "eval_accuracy": 0.4342,
+ "eval_loss": 5.50915888671875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 30976,
+ "train_examples": 45000,
+ "train_loss": 4.405182353210449
+ },
+ {
+ "epoch": 89,
+ "eval_accuracy": 0.4862,
+ "eval_loss": 3.551749169921875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 31328,
+ "train_examples": 45000,
+ "train_loss": 4.707141447533502
+ },
+ {
+ "epoch": 90,
+ "eval_accuracy": 0.4614,
+ "eval_loss": 5.64707607421875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 31680,
+ "train_examples": 45000,
+ "train_loss": 4.449501358328925
+ },
+ {
+ "epoch": 91,
+ "eval_accuracy": 0.4286,
+ "eval_loss": 6.6657060546875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 32032,
+ "train_examples": 45000,
+ "train_loss": 4.62092832438151
+ },
+ {
+ "epoch": 92,
+ "eval_accuracy": 0.391,
+ "eval_loss": 7.563044140625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 32384,
+ "train_examples": 45000,
+ "train_loss": 4.743001677449544
+ },
+ {
+ "epoch": 93,
+ "eval_accuracy": 0.4466,
+ "eval_loss": 6.4824744140625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 32736,
+ "train_examples": 45000,
+ "train_loss": 4.3673409123738605
+ },
+ {
+ "epoch": 94,
+ "eval_accuracy": 0.4352,
+ "eval_loss": 6.52478466796875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 33088,
+ "train_examples": 45000,
+ "train_loss": 4.221223247782389
+ },
+ {
+ "epoch": 95,
+ "eval_accuracy": 0.4942,
+ "eval_loss": 3.797342041015625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 33440,
+ "train_examples": 45000,
+ "train_loss": 4.0709133242713085
+ },
+ {
+ "epoch": 96,
+ "eval_accuracy": 0.3552,
+ "eval_loss": 11.0053232421875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 33792,
+ "train_examples": 45000,
+ "train_loss": 4.388050371297201
+ },
+ {
+ "epoch": 97,
+ "eval_accuracy": 0.4682,
+ "eval_loss": 5.0339734375,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 34144,
+ "train_examples": 45000,
+ "train_loss": 4.638358270899455
+ },
+ {
+ "epoch": 98,
+ "eval_accuracy": 0.4216,
+ "eval_loss": 7.56427587890625,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 34496,
+ "train_examples": 45000,
+ "train_loss": 4.665724361250136
+ },
+ {
+ "epoch": 99,
+ "eval_accuracy": 0.3896,
+ "eval_loss": 5.62162314453125,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 34848,
+ "train_examples": 45000,
+ "train_loss": 4.479719784079657
+ },
+ {
+ "epoch": 100,
+ "eval_accuracy": 0.4674,
+ "eval_loss": 6.46364716796875,
+ "eval_split": "validation",
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 35200,
+ "train_examples": 45000,
+ "train_loss": 5.066503650241428
+ },
+ {
+ "epoch": 101,
+ "eval_accuracy": 0.609,
+ "eval_loss": 1.586387109375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 35552,
+ "train_examples": 45000,
+ "train_loss": 1.721678271569146
+ },
+ {
+ "epoch": 102,
+ "eval_accuracy": 0.5992,
+ "eval_loss": 1.4983593505859376,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 35904,
+ "train_examples": 45000,
+ "train_loss": 1.3279080866071913
+ },
+ {
+ "epoch": 103,
+ "eval_accuracy": 0.6246,
+ "eval_loss": 1.256317919921875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 36256,
+ "train_examples": 45000,
+ "train_loss": 1.232182222451104
+ },
+ {
+ "epoch": 104,
+ "eval_accuracy": 0.5786,
+ "eval_loss": 1.4460140625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 36608,
+ "train_examples": 45000,
+ "train_loss": 1.1727672132703992
+ },
+ {
+ "epoch": 105,
+ "eval_accuracy": 0.6232,
+ "eval_loss": 1.2341310546875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 36960,
+ "train_examples": 45000,
+ "train_loss": 1.1651357496897379
+ },
+ {
+ "epoch": 106,
+ "eval_accuracy": 0.6248,
+ "eval_loss": 1.1990308349609375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 37312,
+ "train_examples": 45000,
+ "train_loss": 1.153296654340956
+ },
+ {
+ "epoch": 107,
+ "eval_accuracy": 0.627,
+ "eval_loss": 1.151527783203125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 37664,
+ "train_examples": 45000,
+ "train_loss": 1.1303095881144205
+ },
+ {
+ "epoch": 108,
+ "eval_accuracy": 0.6266,
+ "eval_loss": 1.139106787109375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 38016,
+ "train_examples": 45000,
+ "train_loss": 1.1244081782446966
+ },
+ {
+ "epoch": 109,
+ "eval_accuracy": 0.6224,
+ "eval_loss": 1.18290263671875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 38368,
+ "train_examples": 45000,
+ "train_loss": 1.121556561088562
+ },
+ {
+ "epoch": 110,
+ "eval_accuracy": 0.623,
+ "eval_loss": 1.1661594482421875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 38720,
+ "train_examples": 45000,
+ "train_loss": 1.1135010125690037
+ },
+ {
+ "epoch": 111,
+ "eval_accuracy": 0.6088,
+ "eval_loss": 1.201440478515625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 39072,
+ "train_examples": 45000,
+ "train_loss": 1.1076628268559774
+ },
+ {
+ "epoch": 112,
+ "eval_accuracy": 0.6354,
+ "eval_loss": 1.09391640625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 39424,
+ "train_examples": 45000,
+ "train_loss": 1.1044298093159994
+ },
+ {
+ "epoch": 113,
+ "eval_accuracy": 0.6282,
+ "eval_loss": 1.1468849365234375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 39776,
+ "train_examples": 45000,
+ "train_loss": 1.090382902208964
+ },
+ {
+ "epoch": 114,
+ "eval_accuracy": 0.6192,
+ "eval_loss": 1.1721736328125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 40128,
+ "train_examples": 45000,
+ "train_loss": 1.0959963248146904
+ },
+ {
+ "epoch": 115,
+ "eval_accuracy": 0.6326,
+ "eval_loss": 1.1188074951171876,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 40480,
+ "train_examples": 45000,
+ "train_loss": 1.0874598748736912
+ },
+ {
+ "epoch": 116,
+ "eval_accuracy": 0.6348,
+ "eval_loss": 1.119938037109375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 40832,
+ "train_examples": 45000,
+ "train_loss": 1.0830459158155654
+ },
+ {
+ "epoch": 117,
+ "eval_accuracy": 0.6334,
+ "eval_loss": 1.0968715087890626,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 41184,
+ "train_examples": 45000,
+ "train_loss": 1.0895018880844116
+ },
+ {
+ "epoch": 118,
+ "eval_accuracy": 0.6556,
+ "eval_loss": 1.06753017578125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 41536,
+ "train_examples": 45000,
+ "train_loss": 1.0812106000264485
+ },
+ {
+ "epoch": 119,
+ "eval_accuracy": 0.6272,
+ "eval_loss": 1.1438458740234374,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 41888,
+ "train_examples": 45000,
+ "train_loss": 1.0666006845792135
+ },
+ {
+ "epoch": 120,
+ "eval_accuracy": 0.6512,
+ "eval_loss": 1.0582649169921876,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 42240,
+ "train_examples": 45000,
+ "train_loss": 1.0607601169480219
+ },
+ {
+ "epoch": 121,
+ "eval_accuracy": 0.6484,
+ "eval_loss": 1.0663829833984375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 42592,
+ "train_examples": 45000,
+ "train_loss": 1.0521053105036418
+ },
+ {
+ "epoch": 122,
+ "eval_accuracy": 0.6476,
+ "eval_loss": 1.0945978271484376,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 42944,
+ "train_examples": 45000,
+ "train_loss": 1.049345637872484
+ },
+ {
+ "epoch": 123,
+ "eval_accuracy": 0.6316,
+ "eval_loss": 1.1688353271484375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 43296,
+ "train_examples": 45000,
+ "train_loss": 1.0518553288565742
+ },
+ {
+ "epoch": 124,
+ "eval_accuracy": 0.6596,
+ "eval_loss": 1.0690121826171874,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 43648,
+ "train_examples": 45000,
+ "train_loss": 1.050257617494795
+ },
+ {
+ "epoch": 125,
+ "eval_accuracy": 0.6264,
+ "eval_loss": 1.1545436279296875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 44000,
+ "train_examples": 45000,
+ "train_loss": 1.0591116312238906
+ },
+ {
+ "epoch": 126,
+ "eval_accuracy": 0.6508,
+ "eval_loss": 1.0498425415039063,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 44352,
+ "train_examples": 45000,
+ "train_loss": 1.0364673189057245
+ },
+ {
+ "epoch": 127,
+ "eval_accuracy": 0.6358,
+ "eval_loss": 1.1038102294921874,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 44704,
+ "train_examples": 45000,
+ "train_loss": 1.04405638639662
+ },
+ {
+ "epoch": 128,
+ "eval_accuracy": 0.6582,
+ "eval_loss": 1.0474168212890624,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 45056,
+ "train_examples": 45000,
+ "train_loss": 1.041512263361613
+ },
+ {
+ "epoch": 129,
+ "eval_accuracy": 0.6444,
+ "eval_loss": 1.0852489501953124,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 45408,
+ "train_examples": 45000,
+ "train_loss": 1.0383492436091104
+ },
+ {
+ "epoch": 130,
+ "eval_accuracy": 0.628,
+ "eval_loss": 1.138535400390625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 45760,
+ "train_examples": 45000,
+ "train_loss": 1.0526941585858662
+ },
+ {
+ "epoch": 131,
+ "eval_accuracy": 0.6438,
+ "eval_loss": 1.08569150390625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 46112,
+ "train_examples": 45000,
+ "train_loss": 1.0424592800352308
+ },
+ {
+ "epoch": 132,
+ "eval_accuracy": 0.6164,
+ "eval_loss": 1.1733404296875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 46464,
+ "train_examples": 45000,
+ "train_loss": 1.0678051187939115
+ },
+ {
+ "epoch": 133,
+ "eval_accuracy": 0.6428,
+ "eval_loss": 1.0911018798828125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 46816,
+ "train_examples": 45000,
+ "train_loss": 1.0511749137242634
+ },
+ {
+ "epoch": 134,
+ "eval_accuracy": 0.6284,
+ "eval_loss": 1.1672166748046875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 47168,
+ "train_examples": 45000,
+ "train_loss": 1.0622696600808037
+ },
+ {
+ "epoch": 135,
+ "eval_accuracy": 0.6302,
+ "eval_loss": 1.107805029296875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 47520,
+ "train_examples": 45000,
+ "train_loss": 1.0528499618742202
+ },
+ {
+ "epoch": 136,
+ "eval_accuracy": 0.6398,
+ "eval_loss": 1.1062733154296875,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 47872,
+ "train_examples": 45000,
+ "train_loss": 1.0609219336191813
+ },
+ {
+ "epoch": 137,
+ "eval_accuracy": 0.6424,
+ "eval_loss": 1.0821095703125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 48224,
+ "train_examples": 45000,
+ "train_loss": 1.0891843078931174
+ },
+ {
+ "epoch": 138,
+ "eval_accuracy": 0.6324,
+ "eval_loss": 1.0882267578125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 48576,
+ "train_examples": 45000,
+ "train_loss": 1.0656020960172017
+ },
+ {
+ "epoch": 139,
+ "eval_accuracy": 0.6506,
+ "eval_loss": 1.062216015625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 48928,
+ "train_examples": 45000,
+ "train_loss": 1.04979617284139
+ },
+ {
+ "epoch": 140,
+ "eval_accuracy": 0.6542,
+ "eval_loss": 1.0492350952148437,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 49280,
+ "train_examples": 45000,
+ "train_loss": 1.039383688206143
+ },
+ {
+ "epoch": 141,
+ "eval_accuracy": 0.634,
+ "eval_loss": 1.0878307861328125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 49632,
+ "train_examples": 45000,
+ "train_loss": 1.046560775396559
+ },
+ {
+ "epoch": 142,
+ "eval_accuracy": 0.6532,
+ "eval_loss": 1.0724321533203125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 49984,
+ "train_examples": 45000,
+ "train_loss": 1.0318034744050768
+ },
+ {
+ "epoch": 143,
+ "eval_accuracy": 0.6452,
+ "eval_loss": 1.06745087890625,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 50336,
+ "train_examples": 45000,
+ "train_loss": 1.0459841170628865
+ },
+ {
+ "epoch": 144,
+ "eval_accuracy": 0.6494,
+ "eval_loss": 1.0546759521484375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 50688,
+ "train_examples": 45000,
+ "train_loss": 1.0411283879597981
+ },
+ {
+ "epoch": 145,
+ "eval_accuracy": 0.6406,
+ "eval_loss": 1.07638974609375,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 51040,
+ "train_examples": 45000,
+ "train_loss": 1.03138457192315
+ },
+ {
+ "epoch": 146,
+ "eval_accuracy": 0.6496,
+ "eval_loss": 1.0701206787109374,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 51392,
+ "train_examples": 45000,
+ "train_loss": 1.021044852172004
+ },
+ {
+ "epoch": 147,
+ "eval_accuracy": 0.66,
+ "eval_loss": 1.0486619995117188,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 51744,
+ "train_examples": 45000,
+ "train_loss": 1.0230277407752144
+ },
+ {
+ "epoch": 148,
+ "eval_accuracy": 0.6444,
+ "eval_loss": 1.066232470703125,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 52096,
+ "train_examples": 45000,
+ "train_loss": 1.0228486086951363
+ },
+ {
+ "epoch": 149,
+ "eval_accuracy": 0.6516,
+ "eval_loss": 1.0704355102539063,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 52448,
+ "train_examples": 45000,
+ "train_loss": 1.0319631608539157
+ },
+ {
+ "epoch": 150,
+ "eval_accuracy": 0.6548,
+ "eval_loss": 1.0434142211914061,
+ "eval_split": "validation",
+ "lr": 0.010000000000000002,
+ "output_lr": 0.010000000000000002,
+ "step": 52800,
+ "train_examples": 45000,
+ "train_loss": 1.0307153160836962
+ },
+ {
+ "epoch": 151,
+ "eval_accuracy": 0.6632,
+ "eval_loss": 1.0081596069335939,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 53152,
+ "train_examples": 45000,
+ "train_loss": 0.968408794424269
+ },
+ {
+ "epoch": 152,
+ "eval_accuracy": 0.6622,
+ "eval_loss": 1.0009671264648436,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 53504,
+ "train_examples": 45000,
+ "train_loss": 0.961383619011773
+ },
+ {
+ "epoch": 153,
+ "eval_accuracy": 0.6626,
+ "eval_loss": 1.0000935791015626,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 53856,
+ "train_examples": 45000,
+ "train_loss": 0.9587891780005561
+ },
+ {
+ "epoch": 154,
+ "eval_accuracy": 0.6666,
+ "eval_loss": 0.9947133178710937,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 54208,
+ "train_examples": 45000,
+ "train_loss": 0.956661511707306
+ },
+ {
+ "epoch": 155,
+ "eval_accuracy": 0.667,
+ "eval_loss": 0.9916768676757812,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 54560,
+ "train_examples": 45000,
+ "train_loss": 0.9527008137491014
+ },
+ {
+ "epoch": 156,
+ "eval_accuracy": 0.6698,
+ "eval_loss": 0.9854952758789063,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 54912,
+ "train_examples": 45000,
+ "train_loss": 0.9535861522886488
+ },
+ {
+ "epoch": 157,
+ "eval_accuracy": 0.6702,
+ "eval_loss": 0.9857482788085937,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 55264,
+ "train_examples": 45000,
+ "train_loss": 0.9495243875291612
+ },
+ {
+ "epoch": 158,
+ "eval_accuracy": 0.6678,
+ "eval_loss": 0.9886399169921875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 55616,
+ "train_examples": 45000,
+ "train_loss": 0.9482226912604438
+ },
+ {
+ "epoch": 159,
+ "eval_accuracy": 0.6692,
+ "eval_loss": 0.9884629272460937,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 55968,
+ "train_examples": 45000,
+ "train_loss": 0.9453291122012668
+ },
+ {
+ "epoch": 160,
+ "eval_accuracy": 0.6702,
+ "eval_loss": 0.98918701171875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 56320,
+ "train_examples": 45000,
+ "train_loss": 0.9501123986562093
+ },
+ {
+ "epoch": 161,
+ "eval_accuracy": 0.6724,
+ "eval_loss": 0.9819580810546875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 56672,
+ "train_examples": 45000,
+ "train_loss": 0.9426417893515693
+ },
+ {
+ "epoch": 162,
+ "eval_accuracy": 0.6688,
+ "eval_loss": 0.9828267333984375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 57024,
+ "train_examples": 45000,
+ "train_loss": 0.9445630306455824
+ },
+ {
+ "epoch": 163,
+ "eval_accuracy": 0.6716,
+ "eval_loss": 0.9865187255859375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 57376,
+ "train_examples": 45000,
+ "train_loss": 0.9460579630851745
+ },
+ {
+ "epoch": 164,
+ "eval_accuracy": 0.672,
+ "eval_loss": 0.9791216430664063,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 57728,
+ "train_examples": 45000,
+ "train_loss": 0.9461332703696357
+ },
+ {
+ "epoch": 165,
+ "eval_accuracy": 0.671,
+ "eval_loss": 0.980871533203125,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 58080,
+ "train_examples": 45000,
+ "train_loss": 0.9464517212549846
+ },
+ {
+ "epoch": 166,
+ "eval_accuracy": 0.6692,
+ "eval_loss": 0.9821738891601562,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 58432,
+ "train_examples": 45000,
+ "train_loss": 0.9537665562417772
+ },
+ {
+ "epoch": 167,
+ "eval_accuracy": 0.6728,
+ "eval_loss": 0.9786509887695313,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 58784,
+ "train_examples": 45000,
+ "train_loss": 0.9519137758678861
+ },
+ {
+ "epoch": 168,
+ "eval_accuracy": 0.6714,
+ "eval_loss": 0.9796623901367187,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 59136,
+ "train_examples": 45000,
+ "train_loss": 0.9474307224697537
+ },
+ {
+ "epoch": 169,
+ "eval_accuracy": 0.6732,
+ "eval_loss": 0.9656408813476562,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 59488,
+ "train_examples": 45000,
+ "train_loss": 0.9520944051954481
+ },
+ {
+ "epoch": 170,
+ "eval_accuracy": 0.6716,
+ "eval_loss": 0.9688844848632813,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 59840,
+ "train_examples": 45000,
+ "train_loss": 0.946247570154402
+ },
+ {
+ "epoch": 171,
+ "eval_accuracy": 0.6714,
+ "eval_loss": 0.9677995239257813,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 60192,
+ "train_examples": 45000,
+ "train_loss": 0.9489625805536905
+ },
+ {
+ "epoch": 172,
+ "eval_accuracy": 0.67,
+ "eval_loss": 0.9761918212890625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 60544,
+ "train_examples": 45000,
+ "train_loss": 0.9467603574222988
+ },
+ {
+ "epoch": 173,
+ "eval_accuracy": 0.6702,
+ "eval_loss": 0.968112744140625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 60896,
+ "train_examples": 45000,
+ "train_loss": 0.9482885099093119
+ },
+ {
+ "epoch": 174,
+ "eval_accuracy": 0.6734,
+ "eval_loss": 0.975332275390625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 61248,
+ "train_examples": 45000,
+ "train_loss": 0.9473493633694119
+ },
+ {
+ "epoch": 175,
+ "eval_accuracy": 0.6762,
+ "eval_loss": 0.9680783447265625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 61600,
+ "train_examples": 45000,
+ "train_loss": 0.9499148143344456
+ },
+ {
+ "epoch": 176,
+ "eval_accuracy": 0.6714,
+ "eval_loss": 0.9696656860351562,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 61952,
+ "train_examples": 45000,
+ "train_loss": 0.9587628659142389
+ },
+ {
+ "epoch": 177,
+ "eval_accuracy": 0.6732,
+ "eval_loss": 0.9686334350585938,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 62304,
+ "train_examples": 45000,
+ "train_loss": 0.955567098681132
+ },
+ {
+ "epoch": 178,
+ "eval_accuracy": 0.6734,
+ "eval_loss": 0.9767455810546875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 62656,
+ "train_examples": 45000,
+ "train_loss": 0.9594909250683255
+ },
+ {
+ "epoch": 179,
+ "eval_accuracy": 0.6736,
+ "eval_loss": 0.9675688354492188,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 63008,
+ "train_examples": 45000,
+ "train_loss": 0.9592544206619262
+ },
+ {
+ "epoch": 180,
+ "eval_accuracy": 0.674,
+ "eval_loss": 0.9692090087890625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 63360,
+ "train_examples": 45000,
+ "train_loss": 0.9600356777615018
+ },
+ {
+ "epoch": 181,
+ "eval_accuracy": 0.673,
+ "eval_loss": 0.9752307861328126,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 63712,
+ "train_examples": 45000,
+ "train_loss": 0.9585328283309936
+ },
+ {
+ "epoch": 182,
+ "eval_accuracy": 0.6732,
+ "eval_loss": 0.9749939697265625,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 64064,
+ "train_examples": 45000,
+ "train_loss": 0.9576578861024645
+ },
+ {
+ "epoch": 183,
+ "eval_accuracy": 0.668,
+ "eval_loss": 0.9789309448242187,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 64416,
+ "train_examples": 45000,
+ "train_loss": 0.9648777356147766
+ },
+ {
+ "epoch": 184,
+ "eval_accuracy": 0.6728,
+ "eval_loss": 0.9707718383789062,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 64768,
+ "train_examples": 45000,
+ "train_loss": 0.9659501979933844
+ },
+ {
+ "epoch": 185,
+ "eval_accuracy": 0.6704,
+ "eval_loss": 0.97909375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 65120,
+ "train_examples": 45000,
+ "train_loss": 0.969595404349433
+ },
+ {
+ "epoch": 186,
+ "eval_accuracy": 0.6708,
+ "eval_loss": 0.9784969604492187,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 65472,
+ "train_examples": 45000,
+ "train_loss": 0.9665438864601983
+ },
+ {
+ "epoch": 187,
+ "eval_accuracy": 0.6696,
+ "eval_loss": 0.981677734375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 65824,
+ "train_examples": 45000,
+ "train_loss": 0.9694981577661302
+ },
+ {
+ "epoch": 188,
+ "eval_accuracy": 0.6668,
+ "eval_loss": 0.9806268432617188,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 66176,
+ "train_examples": 45000,
+ "train_loss": 0.9674224305470784
+ },
+ {
+ "epoch": 189,
+ "eval_accuracy": 0.668,
+ "eval_loss": 0.97913740234375,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 66528,
+ "train_examples": 45000,
+ "train_loss": 0.9666299751281738
+ },
+ {
+ "epoch": 190,
+ "eval_accuracy": 0.6686,
+ "eval_loss": 0.9715440063476563,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 66880,
+ "train_examples": 45000,
+ "train_loss": 0.9722860072347853
+ },
+ {
+ "epoch": 191,
+ "eval_accuracy": 0.6676,
+ "eval_loss": 0.9761428833007812,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 67232,
+ "train_examples": 45000,
+ "train_loss": 0.9745998671107822
+ },
+ {
+ "epoch": 192,
+ "eval_accuracy": 0.6666,
+ "eval_loss": 0.9851692260742188,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 67584,
+ "train_examples": 45000,
+ "train_loss": 0.9772617618454827
+ },
+ {
+ "epoch": 193,
+ "eval_accuracy": 0.6644,
+ "eval_loss": 0.9863299438476563,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 67936,
+ "train_examples": 45000,
+ "train_loss": 0.9811174005296496
+ },
+ {
+ "epoch": 194,
+ "eval_accuracy": 0.6638,
+ "eval_loss": 0.9889223022460938,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 68288,
+ "train_examples": 45000,
+ "train_loss": 0.9881035271220737
+ },
+ {
+ "epoch": 195,
+ "eval_accuracy": 0.6534,
+ "eval_loss": 1.0029588500976563,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 68640,
+ "train_examples": 45000,
+ "train_loss": 0.9892041141933865
+ },
+ {
+ "epoch": 196,
+ "eval_accuracy": 0.649,
+ "eval_loss": 1.0210272705078125,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 68992,
+ "train_examples": 45000,
+ "train_loss": 1.0075765220854018
+ },
+ {
+ "epoch": 197,
+ "eval_accuracy": 0.6484,
+ "eval_loss": 1.0257604370117188,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 69344,
+ "train_examples": 45000,
+ "train_loss": 1.0129783128950332
+ },
+ {
+ "epoch": 198,
+ "eval_accuracy": 0.6512,
+ "eval_loss": 1.0173834716796875,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 69696,
+ "train_examples": 45000,
+ "train_loss": 1.0185926798078748
+ },
+ {
+ "epoch": 199,
+ "eval_accuracy": 0.651,
+ "eval_loss": 1.0110072143554687,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 70048,
+ "train_examples": 45000,
+ "train_loss": 1.0170071312798394
+ },
+ {
+ "epoch": 200,
+ "eval_accuracy": 0.6548,
+ "eval_loss": 0.9980604614257812,
+ "eval_split": "validation",
+ "lr": 0.0010000000000000002,
+ "output_lr": 0.0010000000000000002,
+ "step": 70400,
+ "train_examples": 45000,
+ "train_loss": 1.0106280994839139
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 0,
+ "test_used_for_selection": false,
+ "validation_evaluations": 201
+ },
+ "final": {
+ "accuracy": 0.6548,
+ "epoch": 200,
+ "evaluation_split": "validation",
+ "finite": true,
+ "loss": 0.9980604614257812,
+ "step": 70400
+ },
+ "hardware": {
+ "cuda_device_name": "NVIDIA RTX A6000",
+ "cuda_visible_devices": "0",
+ "device": "cuda",
+ "device_total_memory_bytes": 50896568320,
+ "peak_memory_allocated_bytes": 2324148736,
+ "peak_memory_reserved_bytes": 3162505216,
+ "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": 34.08719086647034,
+ "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": 2176.771654367447,
+ "train_wall_s": 2142.661477804184
+ },
+ "work": {
+ "apical_macs_per_example": 68420224,
+ "causal_scalar_observations": 0,
+ "components": {
+ "apical_projection_macs": 615782016000000,
+ "apical_regression_macs": 0,
+ "bp_reverse_macs_estimate": 0,
+ "kp_reciprocal_correlation_macs": 0,
+ "local_weight_correlation_macs": 619763328000000,
+ "mirror_feedback_prediction_macs": 0,
+ "mirror_local_correlation_macs": 0,
+ "mirror_response_macs": 0,
+ "ordinary_forward_macs": 619763328000000,
+ "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": 9000000,
+ "total_forward_equivalent_examples": 9000000,
+ "total_macs_estimate": 1855308672000000
+ }
+}