{
  "architecture": {
    "adaptive_apical_parameters": 0,
    "adaptive_feedback_parameters": 0,
    "base_width": 16,
    "blocks_per_stage": 9,
    "bn_eps": 1e-05,
    "bn_momentum": 0.1,
    "depth": 56,
    "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
    "fixed_feedback_parameters": 848512,
    "fixed_traffic_coefficients": 0,
    "forward_parameters": 853018,
    "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
      ],
      [
        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
      ],
      [
        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
      ],
      [
        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": 56,
    "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-d56-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": 0.08863549338032801,
    "feedback_forward_cosine": [
      0.0882125198841095,
      0.17185500264167786,
      0.10954485088586807,
      0.2577724754810333,
      0.0751032680273056,
      0.23121197521686554,
      0.030148033052682877,
      0.26019108295440674,
      0.050409719347953796,
      0.19008716940879822,
      0.0399165078997612,
      0.21744656562805176,
      0.0001835389994084835,
      0.1896965205669403,
      0.005662070121616125,
      0.20933006703853607,
      0.01125209592282772,
      0.1695132851600647,
      0.017937814816832542,
      0.1433870494365692,
      0.013061119243502617,
      0.1942385882139206,
      0.005128783639520407,
      0.2021099328994751,
      -0.0006857686676084995,
      0.20494231581687927,
      0.0395386703312397,
      0.1583186686038971,
      0.016033835709095,
      0.18731288611888885,
      0.021174851804971695,
      0.16396485269069672,
      0.030186234042048454,
      0.1645135134458542,
      0.007649475708603859,
      0.15658310055732727,
      0.011635647155344486,
      0.1251221001148224,
      -0.008313348516821861,
      0.15915539860725403,
      -0.0014332830905914307,
      0.1680000275373459,
      0.011308528482913971,
      0.16522729396820068,
      0.006545661482959986,
      0.15889963507652283,
      0.006269207689911127,
      0.17684805393218994,
      0.006641895975917578,
      0.15410152077674866,
      0.01070297509431839,
      0.15132766962051392,
      0.00690346397459507,
      0.13190671801567078,
      0.6286851167678833
    ],
    "feedback_forward_norm_ratio": [
      1.6574243307113647,
      1.1986840963363647,
      1.6578441858291626,
      1.143052101135254,
      2.059192657470703,
      1.446895956993103,
      2.117647409439087,
      1.6426010131835938,
      2.1686482429504395,
      1.708986520767212,
      2.569164752960205,
      2.15853214263916,
      2.5361809730529785,
      1.9963648319244385,
      2.6983582973480225,
      2.1399641036987305,
      3.1610701084136963,
      2.68005633354187,
      2.4815921783447266,
      1.5921815633773804,
      2.099860191345215,
      1.5018749237060547,
      2.329667568206787,
      1.6975750923156738,
      2.184513807296753,
      1.6134381294250488,
      2.444791316986084,
      1.7782717943191528,
      2.475090503692627,
      1.9780645370483398,
      2.562943935394287,
      1.9800993204116821,
      2.487797498703003,
      1.9621620178222656,
      2.9988152980804443,
      2.466535806655884,
      2.4743497371673584,
      1.2880759239196777,
      1.9693315029144287,
      1.1085004806518555,
      1.9485834836959839,
      1.1041178703308105,
      1.955966830253601,
      1.1079226732254028,
      1.9907394647598267,
      1.2147018909454346,
      2.0718092918395996,
      1.319077968597412,
      2.0721075534820557,
      1.3644046783447266,
      2.153334856033325,
      1.4132341146469116,
      2.29006028175354,
      1.3792732954025269,
      0.112843818962574
    ],
    "feedback_forward_relative_error": [
      1.8586674928665161,
      1.4229700565338135,
      1.8398997783660889,
      1.3104478120803833,
      2.220578908920288,
      1.5570577383041382,
      2.3144640922546387,
      1.6862257719039917,
      2.3418779373168945,
      1.8085689544677734,
      2.7194676399230957,
      2.172678232192993,
      2.7260379791259766,
      2.0562262535095215,
      2.872382402420044,
      2.1641461849212646,
      3.3047280311584473,
      2.6970527172088623,
      2.6588101387023926,
      1.7545499801635742,
      2.3139920234680176,
      1.6346817016601562,
      2.5305044651031494,
      1.787615180015564,
      2.4031431674957275,
      1.7151849269866943,
      2.6045494079589844,
      1.8971514701843262,
      2.654562473297119,
      2.0424752235412598,
      2.73132586479187,
      2.06675124168396,
      2.6531007289886475,
      2.0504820346832275,
      3.1538889408111572,
      2.512242555618286,
      2.6579737663269043,
      1.5286616086959839,
      2.2160797119140625,
      1.369644284248352,
      2.1914751529693604,
      1.3594456911087036,
      2.1866798400878906,
      1.3643220663070679,
      2.2219321727752686,
      1.4454996585845947,
      2.2948672771453857,
      1.507784366607666,
      2.294799327850342,
      1.5623975992202759,
      2.36447811126709,
      1.6029685735702515,
      2.4925405979156494,
      1.5932745933532715,
      0.9331920146942139
    ],
    "innovation_negative_gradient_cosine": [
      0.008553816005587578,
      0.025900859385728836,
      0.0376245342195034,
      0.03193657472729683,
      0.09551645815372467,
      0.039293598383665085,
      0.13530397415161133,
      0.04041023924946785,
      0.158193439245224,
      0.035545770078897476,
      0.17993035912513733,
      0.04054297134280205,
      0.19581130146980286,
      0.04374796524643898,
      0.21109670400619507,
      0.04255548119544983,
      0.22164040803909302,
      0.05183442682027817,
      0.224349245429039,
      0.04830259829759598,
      0.26252907514572144,
      0.06813947856426239,
      0.2766053080558777,
      0.07549480348825455,
      0.29249435663223267,
      0.09466566145420074,
      0.3062906861305237,
      0.04542122036218643,
      0.31328457593917847,
      0.07723840326070786,
      0.3224060535430908,
      0.04863288253545761,
      0.327239990234375,
      0.04049643874168396,
      0.3324066400527954,
      0.08639751374721527,
      0.3271884620189667,
      0.06347719579935074,
      0.39523422718048096,
      0.04708479344844818,
      0.4143325090408325,
      0.06797029078006744,
      0.42466670274734497,
      0.08129915595054626,
      0.44150951504707336,
      0.07288248836994171,
      0.44816941022872925,
      0.15117616951465607,
      0.4542672038078308,
      0.14126504957675934,
      0.4803707003593445,
      0.19390016794204712,
      0.4851304888725281,
      0.22901073098182678,
      0.6826860904693604
    ],
    "mean_feedback_forward_cosine": 0.11095394464209676,
    "mean_feedback_forward_relative_error": 2.098663744059476,
    "normalization_state": "training_batch_stats_without_running_update",
    "raw_negative_gradient_cosine": [
      0.008553816005587578,
      0.025900859385728836,
      0.0376245342195034,
      0.03193657472729683,
      0.09551645815372467,
      0.039293598383665085,
      0.13530397415161133,
      0.04041023924946785,
      0.158193439245224,
      0.035545770078897476,
      0.17993035912513733,
      0.04054297134280205,
      0.19581130146980286,
      0.04374796524643898,
      0.21109670400619507,
      0.04255548119544983,
      0.22164040803909302,
      0.05183442682027817,
      0.224349245429039,
      0.04830259829759598,
      0.26252907514572144,
      0.06813947856426239,
      0.2766053080558777,
      0.07549480348825455,
      0.29249435663223267,
      0.09466566145420074,
      0.3062906861305237,
      0.04542122036218643,
      0.31328457593917847,
      0.07723840326070786,
      0.3224060535430908,
      0.04863288253545761,
      0.327239990234375,
      0.04049643874168396,
      0.3324066400527954,
      0.08639751374721527,
      0.3271884620189667,
      0.06347719579935074,
      0.39523422718048096,
      0.04708479344844818,
      0.4143325090408325,
      0.06797029078006744,
      0.42466670274734497,
      0.08129915595054626,
      0.44150951504707336,
      0.07288248836994171,
      0.44816941022872925,
      0.15117616951465607,
      0.4542672038078308,
      0.14126504957675934,
      0.4803707003593445,
      0.19390016794204712,
      0.4851304888725281,
      0.22901073098182678,
      0.6826860904693604
    ],
    "teaching_negative_gradient_cosine": [
      0.008553816005587578,
      0.025900859385728836,
      0.0376245342195034,
      0.03193657472729683,
      0.09551645815372467,
      0.039293598383665085,
      0.13530397415161133,
      0.04041023924946785,
      0.158193439245224,
      0.035545770078897476,
      0.17993035912513733,
      0.04054297134280205,
      0.19581130146980286,
      0.04374796524643898,
      0.21109670400619507,
      0.04255548119544983,
      0.22164040803909302,
      0.05183442682027817,
      0.224349245429039,
      0.04830259829759598,
      0.26252907514572144,
      0.06813947856426239,
      0.2766053080558777,
      0.07549480348825455,
      0.29249435663223267,
      0.09466566145420074,
      0.3062906861305237,
      0.04542122036218643,
      0.31328457593917847,
      0.07723840326070786,
      0.3224060535430908,
      0.04863288253545761,
      0.327239990234375,
      0.04049643874168396,
      0.3324066400527954,
      0.08639751374721527,
      0.3271884620189667,
      0.06347719579935074,
      0.39523422718048096,
      0.04708479344844818,
      0.4143325090408325,
      0.06797029078006744,
      0.42466670274734497,
      0.08129915595054626,
      0.44150951504707336,
      0.07288248836994171,
      0.44816941022872925,
      0.15117616951465607,
      0.4542672038078308,
      0.14126504957675934,
      0.4803707003593445,
      0.19390016794204712,
      0.4851304888725281,
      0.22901073098182678,
      0.6826860904693604
    ],
    "wall_s": 0.5787839889526367
  },
  "epochs": [
    {
      "epoch": 1,
      "eval_accuracy": 0.2598,
      "eval_loss": 2.63627958984375,
      "eval_split": "validation",
      "lr": 0.02,
      "output_lr": 0.02,
      "step": 352,
      "train_examples": 45000,
      "train_loss": 2.7094667250739204
    },
    {
      "epoch": 2,
      "eval_accuracy": 0.244,
      "eval_loss": 8.75294052734375,
      "eval_split": "validation",
      "lr": 0.04,
      "output_lr": 0.04,
      "step": 704,
      "train_examples": 45000,
      "train_loss": 7.502461778089735
    },
    {
      "epoch": 3,
      "eval_accuracy": 0.248,
      "eval_loss": 14.208425,
      "eval_split": "validation",
      "lr": 0.06000000000000001,
      "output_lr": 0.06000000000000001,
      "step": 1056,
      "train_examples": 45000,
      "train_loss": 13.557118966674805
    },
    {
      "epoch": 4,
      "eval_accuracy": 0.2646,
      "eval_loss": 19.696252734375,
      "eval_split": "validation",
      "lr": 0.08,
      "output_lr": 0.08,
      "step": 1408,
      "train_examples": 45000,
      "train_loss": 19.190646782769097
    },
    {
      "epoch": 5,
      "eval_accuracy": 0.2452,
      "eval_loss": 33.2675609375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1760,
      "train_examples": 45000,
      "train_loss": 23.807205302937824
    },
    {
      "epoch": 6,
      "eval_accuracy": 0.2678,
      "eval_loss": 40.026609375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2112,
      "train_examples": 45000,
      "train_loss": 24.055041174994575
    },
    {
      "epoch": 7,
      "eval_accuracy": 0.3196,
      "eval_loss": 23.8587390625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2464,
      "train_examples": 45000,
      "train_loss": 24.712686389160154
    },
    {
      "epoch": 8,
      "eval_accuracy": 0.3278,
      "eval_loss": 27.1260609375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2816,
      "train_examples": 45000,
      "train_loss": 23.086525019327798
    },
    {
      "epoch": 9,
      "eval_accuracy": 0.2696,
      "eval_loss": 30.96389921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3168,
      "train_examples": 45000,
      "train_loss": 24.295748993937174
    },
    {
      "epoch": 10,
      "eval_accuracy": 0.2384,
      "eval_loss": 51.4358140625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3520,
      "train_examples": 45000,
      "train_loss": 21.023040753851998
    },
    {
      "epoch": 11,
      "eval_accuracy": 0.3404,
      "eval_loss": 25.350265234375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3872,
      "train_examples": 45000,
      "train_loss": 23.475594851345488
    },
    {
      "epoch": 12,
      "eval_accuracy": 0.336,
      "eval_loss": 22.1634453125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4224,
      "train_examples": 45000,
      "train_loss": 22.52266331515842
    },
    {
      "epoch": 13,
      "eval_accuracy": 0.3306,
      "eval_loss": 23.82544453125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4576,
      "train_examples": 45000,
      "train_loss": 23.4229757297092
    },
    {
      "epoch": 14,
      "eval_accuracy": 0.2642,
      "eval_loss": 34.5797359375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4928,
      "train_examples": 45000,
      "train_loss": 23.008342339409722
    },
    {
      "epoch": 15,
      "eval_accuracy": 0.375,
      "eval_loss": 23.1807421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5280,
      "train_examples": 45000,
      "train_loss": 20.60292657385932
    },
    {
      "epoch": 16,
      "eval_accuracy": 0.2904,
      "eval_loss": 36.161196875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5632,
      "train_examples": 45000,
      "train_loss": 19.139730751546224
    },
    {
      "epoch": 17,
      "eval_accuracy": 0.3454,
      "eval_loss": 26.833031640625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5984,
      "train_examples": 45000,
      "train_loss": 20.002529979112413
    },
    {
      "epoch": 18,
      "eval_accuracy": 0.3586,
      "eval_loss": 21.24528046875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 6336,
      "train_examples": 45000,
      "train_loss": 18.727348812188044
    },
    {
      "epoch": 19,
      "eval_accuracy": 0.3632,
      "eval_loss": 25.525848828125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 6688,
      "train_examples": 45000,
      "train_loss": 17.44186926167806
    },
    {
      "epoch": 20,
      "eval_accuracy": 0.301,
      "eval_loss": 36.5179703125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7040,
      "train_examples": 45000,
      "train_loss": 17.90213321668837
    },
    {
      "epoch": 21,
      "eval_accuracy": 0.3958,
      "eval_loss": 19.76304296875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7392,
      "train_examples": 45000,
      "train_loss": 17.890375726996528
    },
    {
      "epoch": 22,
      "eval_accuracy": 0.327,
      "eval_loss": 28.7622765625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7744,
      "train_examples": 45000,
      "train_loss": 17.43689203559028
    },
    {
      "epoch": 23,
      "eval_accuracy": 0.2592,
      "eval_loss": 41.68627578125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8096,
      "train_examples": 45000,
      "train_loss": 17.345420552571614
    },
    {
      "epoch": 24,
      "eval_accuracy": 0.4054,
      "eval_loss": 21.936447265625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8448,
      "train_examples": 45000,
      "train_loss": 16.934118476019965
    },
    {
      "epoch": 25,
      "eval_accuracy": 0.2902,
      "eval_loss": 30.93798828125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8800,
      "train_examples": 45000,
      "train_loss": 18.798614492119682
    },
    {
      "epoch": 26,
      "eval_accuracy": 0.3712,
      "eval_loss": 29.504349609375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9152,
      "train_examples": 45000,
      "train_loss": 18.812271053059895
    },
    {
      "epoch": 27,
      "eval_accuracy": 0.342,
      "eval_loss": 26.395713671875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9504,
      "train_examples": 45000,
      "train_loss": 18.473190761650933
    },
    {
      "epoch": 28,
      "eval_accuracy": 0.3854,
      "eval_loss": 26.093215625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9856,
      "train_examples": 45000,
      "train_loss": 18.005538069322373
    },
    {
      "epoch": 29,
      "eval_accuracy": 0.3012,
      "eval_loss": 25.42871171875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10208,
      "train_examples": 45000,
      "train_loss": 18.125954042222766
    },
    {
      "epoch": 30,
      "eval_accuracy": 0.4244,
      "eval_loss": 19.527813671875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10560,
      "train_examples": 45000,
      "train_loss": 18.05612528415256
    },
    {
      "epoch": 31,
      "eval_accuracy": 0.3614,
      "eval_loss": 21.34248828125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10912,
      "train_examples": 45000,
      "train_loss": 19.15135492943658
    },
    {
      "epoch": 32,
      "eval_accuracy": 0.4132,
      "eval_loss": 20.4143234375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11264,
      "train_examples": 45000,
      "train_loss": 18.49009884847005
    },
    {
      "epoch": 33,
      "eval_accuracy": 0.4256,
      "eval_loss": 16.4301474609375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11616,
      "train_examples": 45000,
      "train_loss": 18.47117029859755
    },
    {
      "epoch": 34,
      "eval_accuracy": 0.4202,
      "eval_loss": 15.4893919921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11968,
      "train_examples": 45000,
      "train_loss": 17.859162979804147
    },
    {
      "epoch": 35,
      "eval_accuracy": 0.388,
      "eval_loss": 16.226535546875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12320,
      "train_examples": 45000,
      "train_loss": 18.110402368503145
    },
    {
      "epoch": 36,
      "eval_accuracy": 0.4052,
      "eval_loss": 18.623937890625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12672,
      "train_examples": 45000,
      "train_loss": 17.06648885972765
    },
    {
      "epoch": 37,
      "eval_accuracy": 0.3954,
      "eval_loss": 18.58285390625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13024,
      "train_examples": 45000,
      "train_loss": 19.626207305230036
    },
    {
      "epoch": 38,
      "eval_accuracy": 0.424,
      "eval_loss": 26.581407421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13376,
      "train_examples": 45000,
      "train_loss": 18.639922721354168
    },
    {
      "epoch": 39,
      "eval_accuracy": 0.4192,
      "eval_loss": 19.451832421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13728,
      "train_examples": 45000,
      "train_loss": 18.782825440131294
    },
    {
      "epoch": 40,
      "eval_accuracy": 0.3804,
      "eval_loss": 26.744866015625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14080,
      "train_examples": 45000,
      "train_loss": 17.683322509087457
    },
    {
      "epoch": 41,
      "eval_accuracy": 0.395,
      "eval_loss": 34.78169375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14432,
      "train_examples": 45000,
      "train_loss": 18.225805063544378
    },
    {
      "epoch": 42,
      "eval_accuracy": 0.412,
      "eval_loss": 29.765325,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14784,
      "train_examples": 45000,
      "train_loss": 17.984680138481988
    },
    {
      "epoch": 43,
      "eval_accuracy": 0.3786,
      "eval_loss": 19.825704296875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15136,
      "train_examples": 45000,
      "train_loss": 17.825573151312934
    },
    {
      "epoch": 44,
      "eval_accuracy": 0.3604,
      "eval_loss": 26.8407375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15488,
      "train_examples": 45000,
      "train_loss": 19.19038238796658
    },
    {
      "epoch": 45,
      "eval_accuracy": 0.4052,
      "eval_loss": 12.7454291015625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15840,
      "train_examples": 45000,
      "train_loss": 16.198456757269966
    },
    {
      "epoch": 46,
      "eval_accuracy": 0.424,
      "eval_loss": 15.499886328125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16192,
      "train_examples": 45000,
      "train_loss": 16.85351919962565
    },
    {
      "epoch": 47,
      "eval_accuracy": 0.4306,
      "eval_loss": 16.8180568359375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16544,
      "train_examples": 45000,
      "train_loss": 17.23167397986518
    },
    {
      "epoch": 48,
      "eval_accuracy": 0.4574,
      "eval_loss": 15.12474765625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16896,
      "train_examples": 45000,
      "train_loss": 17.16782407752143
    },
    {
      "epoch": 49,
      "eval_accuracy": 0.3906,
      "eval_loss": 33.472205859375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17248,
      "train_examples": 45000,
      "train_loss": 17.989877756754556
    },
    {
      "epoch": 50,
      "eval_accuracy": 0.3944,
      "eval_loss": 18.846742578125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17600,
      "train_examples": 45000,
      "train_loss": 16.750614368184408
    },
    {
      "epoch": 51,
      "eval_accuracy": 0.3306,
      "eval_loss": 20.342544921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17952,
      "train_examples": 45000,
      "train_loss": 16.608609683566623
    },
    {
      "epoch": 52,
      "eval_accuracy": 0.4454,
      "eval_loss": 21.5571515625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 18304,
      "train_examples": 45000,
      "train_loss": 16.961722309366863
    },
    {
      "epoch": 53,
      "eval_accuracy": 0.3998,
      "eval_loss": 23.065123828125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 18656,
      "train_examples": 45000,
      "train_loss": 16.919527493286132
    },
    {
      "epoch": 54,
      "eval_accuracy": 0.3482,
      "eval_loss": 25.1215953125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19008,
      "train_examples": 45000,
      "train_loss": 16.035512492031522
    },
    {
      "epoch": 55,
      "eval_accuracy": 0.469,
      "eval_loss": 15.480585546875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19360,
      "train_examples": 45000,
      "train_loss": 17.075127189466688
    },
    {
      "epoch": 56,
      "eval_accuracy": 0.4464,
      "eval_loss": 14.592352734375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19712,
      "train_examples": 45000,
      "train_loss": 17.66614246198866
    },
    {
      "epoch": 57,
      "eval_accuracy": 0.4314,
      "eval_loss": 22.738269140625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20064,
      "train_examples": 45000,
      "train_loss": 16.894754077487523
    },
    {
      "epoch": 58,
      "eval_accuracy": 0.3318,
      "eval_loss": 45.84163984375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20416,
      "train_examples": 45000,
      "train_loss": 19.05143542683919
    },
    {
      "epoch": 59,
      "eval_accuracy": 0.372,
      "eval_loss": 31.423384375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20768,
      "train_examples": 45000,
      "train_loss": 18.783138044230142
    },
    {
      "epoch": 60,
      "eval_accuracy": 0.418,
      "eval_loss": 16.9460552734375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21120,
      "train_examples": 45000,
      "train_loss": 16.77694915907118
    },
    {
      "epoch": 61,
      "eval_accuracy": 0.361,
      "eval_loss": 22.063344140625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21472,
      "train_examples": 45000,
      "train_loss": 16.14378400200738
    },
    {
      "epoch": 62,
      "eval_accuracy": 0.408,
      "eval_loss": 24.965109765625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21824,
      "train_examples": 45000,
      "train_loss": 17.721834357367623
    },
    {
      "epoch": 63,
      "eval_accuracy": 0.4464,
      "eval_loss": 20.872457421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22176,
      "train_examples": 45000,
      "train_loss": 17.106755812411837
    },
    {
      "epoch": 64,
      "eval_accuracy": 0.351,
      "eval_loss": 27.144359765625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22528,
      "train_examples": 45000,
      "train_loss": 16.525339327663847
    },
    {
      "epoch": 65,
      "eval_accuracy": 0.401,
      "eval_loss": 28.997582421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22880,
      "train_examples": 45000,
      "train_loss": 17.898866891988117
    },
    {
      "epoch": 66,
      "eval_accuracy": 0.4388,
      "eval_loss": 25.025003515625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23232,
      "train_examples": 45000,
      "train_loss": 18.09744178805881
    },
    {
      "epoch": 67,
      "eval_accuracy": 0.4728,
      "eval_loss": 16.6167294921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23584,
      "train_examples": 45000,
      "train_loss": 17.755582858784994
    },
    {
      "epoch": 68,
      "eval_accuracy": 0.4108,
      "eval_loss": 19.801496484375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23936,
      "train_examples": 45000,
      "train_loss": 17.263119559054903
    },
    {
      "epoch": 69,
      "eval_accuracy": 0.4906,
      "eval_loss": 18.20902421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24288,
      "train_examples": 45000,
      "train_loss": 17.33434205186632
    },
    {
      "epoch": 70,
      "eval_accuracy": 0.4292,
      "eval_loss": 34.2252921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24640,
      "train_examples": 45000,
      "train_loss": 18.17401082695855
    },
    {
      "epoch": 71,
      "eval_accuracy": 0.4798,
      "eval_loss": 15.21786796875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24992,
      "train_examples": 45000,
      "train_loss": 18.021223406304255
    },
    {
      "epoch": 72,
      "eval_accuracy": 0.4518,
      "eval_loss": 22.791141015625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25344,
      "train_examples": 45000,
      "train_loss": 19.531677550930446
    },
    {
      "epoch": 73,
      "eval_accuracy": 0.368,
      "eval_loss": 38.81344453125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25696,
      "train_examples": 45000,
      "train_loss": 18.620982016838923
    },
    {
      "epoch": 74,
      "eval_accuracy": 0.4228,
      "eval_loss": 30.719596484375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26048,
      "train_examples": 45000,
      "train_loss": 20.337078319295248
    },
    {
      "epoch": 75,
      "eval_accuracy": 0.4418,
      "eval_loss": 21.886297265625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26400,
      "train_examples": 45000,
      "train_loss": 20.1115397922092
    },
    {
      "epoch": 76,
      "eval_accuracy": 0.4868,
      "eval_loss": 16.3975734375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26752,
      "train_examples": 45000,
      "train_loss": 17.937315833706325
    },
    {
      "epoch": 77,
      "eval_accuracy": 0.325,
      "eval_loss": 45.389496875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27104,
      "train_examples": 45000,
      "train_loss": 21.48694476860894
    },
    {
      "epoch": 78,
      "eval_accuracy": 0.4012,
      "eval_loss": 26.195025390625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27456,
      "train_examples": 45000,
      "train_loss": 22.31231401570638
    },
    {
      "epoch": 79,
      "eval_accuracy": 0.445,
      "eval_loss": 27.56462890625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27808,
      "train_examples": 45000,
      "train_loss": 21.119349666341147
    },
    {
      "epoch": 80,
      "eval_accuracy": 0.471,
      "eval_loss": 17.123298046875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28160,
      "train_examples": 45000,
      "train_loss": 20.679742971123588
    },
    {
      "epoch": 81,
      "eval_accuracy": 0.4214,
      "eval_loss": 27.093696484375,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28512,
      "train_examples": 45000,
      "train_loss": 20.63988988172743
    },
    {
      "epoch": 82,
      "eval_accuracy": 0.4414,
      "eval_loss": 26.012460546875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28864,
      "train_examples": 45000,
      "train_loss": 21.493989885457356
    },
    {
      "epoch": 83,
      "eval_accuracy": 0.3698,
      "eval_loss": 30.0991921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29216,
      "train_examples": 45000,
      "train_loss": 22.169163477918836
    },
    {
      "epoch": 84,
      "eval_accuracy": 0.3986,
      "eval_loss": 31.00485078125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29568,
      "train_examples": 45000,
      "train_loss": 23.3011970703125
    },
    {
      "epoch": 85,
      "eval_accuracy": 0.442,
      "eval_loss": 23.574222265625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29920,
      "train_examples": 45000,
      "train_loss": 21.34971837056478
    },
    {
      "epoch": 86,
      "eval_accuracy": 0.4296,
      "eval_loss": 29.412936328125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30272,
      "train_examples": 45000,
      "train_loss": 21.91920473361545
    },
    {
      "epoch": 87,
      "eval_accuracy": 0.4386,
      "eval_loss": 20.903475390625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30624,
      "train_examples": 45000,
      "train_loss": 22.060477025689018
    },
    {
      "epoch": 88,
      "eval_accuracy": 0.4506,
      "eval_loss": 22.551448046875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30976,
      "train_examples": 45000,
      "train_loss": 21.711965223524306
    },
    {
      "epoch": 89,
      "eval_accuracy": 0.3526,
      "eval_loss": 61.11854921875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 31328,
      "train_examples": 45000,
      "train_loss": 21.463106849839953
    },
    {
      "epoch": 90,
      "eval_accuracy": 0.4564,
      "eval_loss": 23.186648046875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 31680,
      "train_examples": 45000,
      "train_loss": 24.00905051337348
    },
    {
      "epoch": 91,
      "eval_accuracy": 0.376,
      "eval_loss": 60.64270546875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32032,
      "train_examples": 45000,
      "train_loss": 21.28097244635688
    },
    {
      "epoch": 92,
      "eval_accuracy": 0.3868,
      "eval_loss": 27.0673390625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32384,
      "train_examples": 45000,
      "train_loss": 23.36898587782118
    },
    {
      "epoch": 93,
      "eval_accuracy": 0.4672,
      "eval_loss": 22.039832421875,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32736,
      "train_examples": 45000,
      "train_loss": 20.88351126539442
    },
    {
      "epoch": 94,
      "eval_accuracy": 0.3888,
      "eval_loss": 25.698695703125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33088,
      "train_examples": 45000,
      "train_loss": 21.979886994425456
    },
    {
      "epoch": 95,
      "eval_accuracy": 0.4346,
      "eval_loss": 26.78786640625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33440,
      "train_examples": 45000,
      "train_loss": 21.236860284084745
    },
    {
      "epoch": 96,
      "eval_accuracy": 0.3624,
      "eval_loss": 36.5288453125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33792,
      "train_examples": 45000,
      "train_loss": 21.10884918484158
    },
    {
      "epoch": 97,
      "eval_accuracy": 0.4434,
      "eval_loss": 23.964403125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34144,
      "train_examples": 45000,
      "train_loss": 20.604113126966688
    },
    {
      "epoch": 98,
      "eval_accuracy": 0.4674,
      "eval_loss": 24.196286328125,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34496,
      "train_examples": 45000,
      "train_loss": 19.82856385260688
    },
    {
      "epoch": 99,
      "eval_accuracy": 0.4354,
      "eval_loss": 24.922025,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34848,
      "train_examples": 45000,
      "train_loss": 19.128273210313584
    },
    {
      "epoch": 100,
      "eval_accuracy": 0.3788,
      "eval_loss": 31.4631015625,
      "eval_split": "validation",
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35200,
      "train_examples": 45000,
      "train_loss": 21.582442571343314
    },
    {
      "epoch": 101,
      "eval_accuracy": 0.5972,
      "eval_loss": 4.836526171875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 35552,
      "train_examples": 45000,
      "train_loss": 5.9259513893975155
    },
    {
      "epoch": 102,
      "eval_accuracy": 0.5624,
      "eval_loss": 4.101729345703125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 35904,
      "train_examples": 45000,
      "train_loss": 3.619677195019192
    },
    {
      "epoch": 103,
      "eval_accuracy": 0.5986,
      "eval_loss": 2.667718505859375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 36256,
      "train_examples": 45000,
      "train_loss": 2.880115439690484
    },
    {
      "epoch": 104,
      "eval_accuracy": 0.5958,
      "eval_loss": 2.390951025390625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 36608,
      "train_examples": 45000,
      "train_loss": 2.4437281640370685
    },
    {
      "epoch": 105,
      "eval_accuracy": 0.574,
      "eval_loss": 2.260222998046875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 36960,
      "train_examples": 45000,
      "train_loss": 2.226529400253296
    },
    {
      "epoch": 106,
      "eval_accuracy": 0.5504,
      "eval_loss": 2.49472744140625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 37312,
      "train_examples": 45000,
      "train_loss": 2.185105883640713
    },
    {
      "epoch": 107,
      "eval_accuracy": 0.5266,
      "eval_loss": 2.41015546875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 37664,
      "train_examples": 45000,
      "train_loss": 2.0920644403245716
    },
    {
      "epoch": 108,
      "eval_accuracy": 0.5886,
      "eval_loss": 1.7911347412109375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 38016,
      "train_examples": 45000,
      "train_loss": 2.0197150400797526
    },
    {
      "epoch": 109,
      "eval_accuracy": 0.5726,
      "eval_loss": 2.175161474609375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 38368,
      "train_examples": 45000,
      "train_loss": 1.8753828492058648
    },
    {
      "epoch": 110,
      "eval_accuracy": 0.58,
      "eval_loss": 1.9253150146484375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 38720,
      "train_examples": 45000,
      "train_loss": 1.8506183098475137
    },
    {
      "epoch": 111,
      "eval_accuracy": 0.5912,
      "eval_loss": 1.810534423828125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39072,
      "train_examples": 45000,
      "train_loss": 1.7877488494873046
    },
    {
      "epoch": 112,
      "eval_accuracy": 0.5956,
      "eval_loss": 1.6881039306640624,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39424,
      "train_examples": 45000,
      "train_loss": 1.7687504001193577
    },
    {
      "epoch": 113,
      "eval_accuracy": 0.5764,
      "eval_loss": 1.8380051513671876,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39776,
      "train_examples": 45000,
      "train_loss": 1.7672821660783555
    },
    {
      "epoch": 114,
      "eval_accuracy": 0.5794,
      "eval_loss": 1.6862110595703126,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40128,
      "train_examples": 45000,
      "train_loss": 1.6414273686091105
    },
    {
      "epoch": 115,
      "eval_accuracy": 0.5752,
      "eval_loss": 1.6944744384765624,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40480,
      "train_examples": 45000,
      "train_loss": 1.6939214492797852
    },
    {
      "epoch": 116,
      "eval_accuracy": 0.5926,
      "eval_loss": 1.582383349609375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40832,
      "train_examples": 45000,
      "train_loss": 1.597629587088691
    },
    {
      "epoch": 117,
      "eval_accuracy": 0.5588,
      "eval_loss": 1.856890869140625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41184,
      "train_examples": 45000,
      "train_loss": 1.6295654713312784
    },
    {
      "epoch": 118,
      "eval_accuracy": 0.5834,
      "eval_loss": 1.705884130859375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41536,
      "train_examples": 45000,
      "train_loss": 1.5944301096810234
    },
    {
      "epoch": 119,
      "eval_accuracy": 0.608,
      "eval_loss": 1.561981005859375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41888,
      "train_examples": 45000,
      "train_loss": 1.5895581109788683
    },
    {
      "epoch": 120,
      "eval_accuracy": 0.5932,
      "eval_loss": 1.6351948486328125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42240,
      "train_examples": 45000,
      "train_loss": 1.5375917362001208
    },
    {
      "epoch": 121,
      "eval_accuracy": 0.5716,
      "eval_loss": 1.6493072265625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42592,
      "train_examples": 45000,
      "train_loss": 1.5845890295876397
    },
    {
      "epoch": 122,
      "eval_accuracy": 0.6058,
      "eval_loss": 1.49820810546875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42944,
      "train_examples": 45000,
      "train_loss": 1.5254730051252576
    },
    {
      "epoch": 123,
      "eval_accuracy": 0.5652,
      "eval_loss": 1.77642001953125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43296,
      "train_examples": 45000,
      "train_loss": 1.5247097325219048
    },
    {
      "epoch": 124,
      "eval_accuracy": 0.6068,
      "eval_loss": 1.376882470703125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43648,
      "train_examples": 45000,
      "train_loss": 1.5499936188379924
    },
    {
      "epoch": 125,
      "eval_accuracy": 0.5676,
      "eval_loss": 1.7324075927734375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44000,
      "train_examples": 45000,
      "train_loss": 1.615310203382704
    },
    {
      "epoch": 126,
      "eval_accuracy": 0.5954,
      "eval_loss": 1.5954062255859376,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44352,
      "train_examples": 45000,
      "train_loss": 1.4982189151763916
    },
    {
      "epoch": 127,
      "eval_accuracy": 0.5846,
      "eval_loss": 1.7751647705078124,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44704,
      "train_examples": 45000,
      "train_loss": 1.5301205956353081
    },
    {
      "epoch": 128,
      "eval_accuracy": 0.5602,
      "eval_loss": 1.6907564697265625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45056,
      "train_examples": 45000,
      "train_loss": 1.4734539789411756
    },
    {
      "epoch": 129,
      "eval_accuracy": 0.6002,
      "eval_loss": 1.40965439453125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45408,
      "train_examples": 45000,
      "train_loss": 1.4641339381747775
    },
    {
      "epoch": 130,
      "eval_accuracy": 0.6192,
      "eval_loss": 1.4004435791015626,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45760,
      "train_examples": 45000,
      "train_loss": 1.5231939977010092
    },
    {
      "epoch": 131,
      "eval_accuracy": 0.5982,
      "eval_loss": 1.45526376953125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46112,
      "train_examples": 45000,
      "train_loss": 1.497420593600803
    },
    {
      "epoch": 132,
      "eval_accuracy": 0.5884,
      "eval_loss": 1.550844140625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46464,
      "train_examples": 45000,
      "train_loss": 1.469409120665656
    },
    {
      "epoch": 133,
      "eval_accuracy": 0.6256,
      "eval_loss": 1.267538330078125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46816,
      "train_examples": 45000,
      "train_loss": 1.4565586593204074
    },
    {
      "epoch": 134,
      "eval_accuracy": 0.5592,
      "eval_loss": 1.77311513671875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47168,
      "train_examples": 45000,
      "train_loss": 1.437351505109999
    },
    {
      "epoch": 135,
      "eval_accuracy": 0.5574,
      "eval_loss": 1.940101904296875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47520,
      "train_examples": 45000,
      "train_loss": 1.4218262961917454
    },
    {
      "epoch": 136,
      "eval_accuracy": 0.5774,
      "eval_loss": 1.6147098388671874,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47872,
      "train_examples": 45000,
      "train_loss": 1.411111260647244
    },
    {
      "epoch": 137,
      "eval_accuracy": 0.6128,
      "eval_loss": 1.357390869140625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48224,
      "train_examples": 45000,
      "train_loss": 1.5000552250120376
    },
    {
      "epoch": 138,
      "eval_accuracy": 0.5848,
      "eval_loss": 1.632561083984375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48576,
      "train_examples": 45000,
      "train_loss": 1.3685510100682576
    },
    {
      "epoch": 139,
      "eval_accuracy": 0.6142,
      "eval_loss": 1.3133041015625,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48928,
      "train_examples": 45000,
      "train_loss": 1.3762463880327014
    },
    {
      "epoch": 140,
      "eval_accuracy": 0.5754,
      "eval_loss": 1.5640064453125,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49280,
      "train_examples": 45000,
      "train_loss": 1.401551789580451
    },
    {
      "epoch": 141,
      "eval_accuracy": 0.594,
      "eval_loss": 1.479112841796875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49632,
      "train_examples": 45000,
      "train_loss": 1.4058031635072497
    },
    {
      "epoch": 142,
      "eval_accuracy": 0.5914,
      "eval_loss": 1.5028365478515624,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49984,
      "train_examples": 45000,
      "train_loss": 1.4034333087073432
    },
    {
      "epoch": 143,
      "eval_accuracy": 0.6102,
      "eval_loss": 1.5048456298828126,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50336,
      "train_examples": 45000,
      "train_loss": 1.3954313831117418
    },
    {
      "epoch": 144,
      "eval_accuracy": 0.5666,
      "eval_loss": 1.6636828369140626,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50688,
      "train_examples": 45000,
      "train_loss": 1.3667300954606798
    },
    {
      "epoch": 145,
      "eval_accuracy": 0.5612,
      "eval_loss": 1.6101176025390624,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51040,
      "train_examples": 45000,
      "train_loss": 1.404260525724623
    },
    {
      "epoch": 146,
      "eval_accuracy": 0.5556,
      "eval_loss": 1.876030859375,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51392,
      "train_examples": 45000,
      "train_loss": 1.4051593517515395
    },
    {
      "epoch": 147,
      "eval_accuracy": 0.5764,
      "eval_loss": 1.6018065673828126,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51744,
      "train_examples": 45000,
      "train_loss": 1.4870435859256321
    },
    {
      "epoch": 148,
      "eval_accuracy": 0.6056,
      "eval_loss": 1.43818701171875,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52096,
      "train_examples": 45000,
      "train_loss": 1.367290024079217
    },
    {
      "epoch": 149,
      "eval_accuracy": 0.6162,
      "eval_loss": 1.3064309814453126,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52448,
      "train_examples": 45000,
      "train_loss": 1.3197986438751221
    },
    {
      "epoch": 150,
      "eval_accuracy": 0.6296,
      "eval_loss": 1.2498236572265624,
      "eval_split": "validation",
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52800,
      "train_examples": 45000,
      "train_loss": 1.3129838515387642
    },
    {
      "epoch": 151,
      "eval_accuracy": 0.6534,
      "eval_loss": 1.07179833984375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 53152,
      "train_examples": 45000,
      "train_loss": 1.026788145626916
    },
    {
      "epoch": 152,
      "eval_accuracy": 0.6568,
      "eval_loss": 1.0430185913085936,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 53504,
      "train_examples": 45000,
      "train_loss": 1.009344142680698
    },
    {
      "epoch": 153,
      "eval_accuracy": 0.658,
      "eval_loss": 1.0307321044921876,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 53856,
      "train_examples": 45000,
      "train_loss": 1.0141865419387817
    },
    {
      "epoch": 154,
      "eval_accuracy": 0.659,
      "eval_loss": 1.02932734375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 54208,
      "train_examples": 45000,
      "train_loss": 1.0022637588077121
    },
    {
      "epoch": 155,
      "eval_accuracy": 0.6604,
      "eval_loss": 1.0242865844726563,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 54560,
      "train_examples": 45000,
      "train_loss": 0.9988587782012092
    },
    {
      "epoch": 156,
      "eval_accuracy": 0.6674,
      "eval_loss": 1.0057396484375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 54912,
      "train_examples": 45000,
      "train_loss": 0.9961366279178195
    },
    {
      "epoch": 157,
      "eval_accuracy": 0.6548,
      "eval_loss": 1.03570576171875,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 55264,
      "train_examples": 45000,
      "train_loss": 0.9966530807283189
    },
    {
      "epoch": 158,
      "eval_accuracy": 0.6616,
      "eval_loss": 1.0033466552734376,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 55616,
      "train_examples": 45000,
      "train_loss": 1.0007755590226914
    },
    {
      "epoch": 159,
      "eval_accuracy": 0.6624,
      "eval_loss": 0.99225693359375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 55968,
      "train_examples": 45000,
      "train_loss": 1.0026236204147339
    },
    {
      "epoch": 160,
      "eval_accuracy": 0.661,
      "eval_loss": 1.00777568359375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 56320,
      "train_examples": 45000,
      "train_loss": 1.0010402308146158
    },
    {
      "epoch": 161,
      "eval_accuracy": 0.6566,
      "eval_loss": 1.0218382690429688,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 56672,
      "train_examples": 45000,
      "train_loss": 0.9932583589553833
    },
    {
      "epoch": 162,
      "eval_accuracy": 0.6546,
      "eval_loss": 1.0098359619140624,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 57024,
      "train_examples": 45000,
      "train_loss": 1.0004163008266025
    },
    {
      "epoch": 163,
      "eval_accuracy": 0.662,
      "eval_loss": 1.0110062622070313,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 57376,
      "train_examples": 45000,
      "train_loss": 1.0034730763753255
    },
    {
      "epoch": 164,
      "eval_accuracy": 0.6494,
      "eval_loss": 1.0401352172851563,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 57728,
      "train_examples": 45000,
      "train_loss": 1.0073212409337362
    },
    {
      "epoch": 165,
      "eval_accuracy": 0.6474,
      "eval_loss": 1.0578818725585937,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 58080,
      "train_examples": 45000,
      "train_loss": 1.0492681529257033
    },
    {
      "epoch": 166,
      "eval_accuracy": 0.6434,
      "eval_loss": 1.0694183837890625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 58432,
      "train_examples": 45000,
      "train_loss": 1.0751917953279284
    },
    {
      "epoch": 167,
      "eval_accuracy": 0.6294,
      "eval_loss": 1.1221494384765625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 58784,
      "train_examples": 45000,
      "train_loss": 1.0811405420939129
    },
    {
      "epoch": 168,
      "eval_accuracy": 0.6354,
      "eval_loss": 1.0892340576171875,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59136,
      "train_examples": 45000,
      "train_loss": 1.0801018522050645
    },
    {
      "epoch": 169,
      "eval_accuracy": 0.6412,
      "eval_loss": 1.0978382080078124,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59488,
      "train_examples": 45000,
      "train_loss": 1.0860861628214518
    },
    {
      "epoch": 170,
      "eval_accuracy": 0.6346,
      "eval_loss": 1.090638720703125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59840,
      "train_examples": 45000,
      "train_loss": 1.0723167603810628
    },
    {
      "epoch": 171,
      "eval_accuracy": 0.6272,
      "eval_loss": 1.1234599609375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60192,
      "train_examples": 45000,
      "train_loss": 1.0920719518237645
    },
    {
      "epoch": 172,
      "eval_accuracy": 0.6284,
      "eval_loss": 1.099410595703125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60544,
      "train_examples": 45000,
      "train_loss": 1.0925795223236083
    },
    {
      "epoch": 173,
      "eval_accuracy": 0.63,
      "eval_loss": 1.104658056640625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60896,
      "train_examples": 45000,
      "train_loss": 1.087668931304084
    },
    {
      "epoch": 174,
      "eval_accuracy": 0.628,
      "eval_loss": 1.1068023193359375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61248,
      "train_examples": 45000,
      "train_loss": 1.094618272082011
    },
    {
      "epoch": 175,
      "eval_accuracy": 0.6318,
      "eval_loss": 1.09778740234375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61600,
      "train_examples": 45000,
      "train_loss": 1.0823923013687133
    },
    {
      "epoch": 176,
      "eval_accuracy": 0.6308,
      "eval_loss": 1.10460810546875,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61952,
      "train_examples": 45000,
      "train_loss": 1.0776995142830743
    },
    {
      "epoch": 177,
      "eval_accuracy": 0.6418,
      "eval_loss": 1.0786893798828125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62304,
      "train_examples": 45000,
      "train_loss": 1.069732749218411
    },
    {
      "epoch": 178,
      "eval_accuracy": 0.6406,
      "eval_loss": 1.0766607421875,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62656,
      "train_examples": 45000,
      "train_loss": 1.0717433659447564
    },
    {
      "epoch": 179,
      "eval_accuracy": 0.635,
      "eval_loss": 1.0887271728515624,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63008,
      "train_examples": 45000,
      "train_loss": 1.0777871327506172
    },
    {
      "epoch": 180,
      "eval_accuracy": 0.6436,
      "eval_loss": 1.062487890625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63360,
      "train_examples": 45000,
      "train_loss": 1.070056160566542
    },
    {
      "epoch": 181,
      "eval_accuracy": 0.6448,
      "eval_loss": 1.060628564453125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63712,
      "train_examples": 45000,
      "train_loss": 1.0583967365688747
    },
    {
      "epoch": 182,
      "eval_accuracy": 0.6446,
      "eval_loss": 1.058402197265625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64064,
      "train_examples": 45000,
      "train_loss": 1.0691630625618829
    },
    {
      "epoch": 183,
      "eval_accuracy": 0.6366,
      "eval_loss": 1.0809754150390625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64416,
      "train_examples": 45000,
      "train_loss": 1.0663337293412951
    },
    {
      "epoch": 184,
      "eval_accuracy": 0.6452,
      "eval_loss": 1.058856298828125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64768,
      "train_examples": 45000,
      "train_loss": 1.0665179736455281
    },
    {
      "epoch": 185,
      "eval_accuracy": 0.6398,
      "eval_loss": 1.06523369140625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65120,
      "train_examples": 45000,
      "train_loss": 1.0648094150119358
    },
    {
      "epoch": 186,
      "eval_accuracy": 0.64,
      "eval_loss": 1.060458837890625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65472,
      "train_examples": 45000,
      "train_loss": 1.0582156455569798
    },
    {
      "epoch": 187,
      "eval_accuracy": 0.6538,
      "eval_loss": 1.04734462890625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65824,
      "train_examples": 45000,
      "train_loss": 1.0498132875124613
    },
    {
      "epoch": 188,
      "eval_accuracy": 0.6562,
      "eval_loss": 1.0362562622070313,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66176,
      "train_examples": 45000,
      "train_loss": 1.0539691292444864
    },
    {
      "epoch": 189,
      "eval_accuracy": 0.6546,
      "eval_loss": 1.0296682861328126,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66528,
      "train_examples": 45000,
      "train_loss": 1.050129378912184
    },
    {
      "epoch": 190,
      "eval_accuracy": 0.6484,
      "eval_loss": 1.0466437377929687,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66880,
      "train_examples": 45000,
      "train_loss": 1.051998159429762
    },
    {
      "epoch": 191,
      "eval_accuracy": 0.6538,
      "eval_loss": 1.0468611083984376,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67232,
      "train_examples": 45000,
      "train_loss": 1.0416840789159139
    },
    {
      "epoch": 192,
      "eval_accuracy": 0.6592,
      "eval_loss": 1.0471241577148438,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67584,
      "train_examples": 45000,
      "train_loss": 1.0441420639038086
    },
    {
      "epoch": 193,
      "eval_accuracy": 0.6478,
      "eval_loss": 1.0665528564453124,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67936,
      "train_examples": 45000,
      "train_loss": 1.048444183370802
    },
    {
      "epoch": 194,
      "eval_accuracy": 0.647,
      "eval_loss": 1.0716345703125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68288,
      "train_examples": 45000,
      "train_loss": 1.0648873981157938
    },
    {
      "epoch": 195,
      "eval_accuracy": 0.6476,
      "eval_loss": 1.0518780395507812,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68640,
      "train_examples": 45000,
      "train_loss": 1.0581021981451246
    },
    {
      "epoch": 196,
      "eval_accuracy": 0.647,
      "eval_loss": 1.0528640258789062,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68992,
      "train_examples": 45000,
      "train_loss": 1.0561888460053337
    },
    {
      "epoch": 197,
      "eval_accuracy": 0.6446,
      "eval_loss": 1.0466599609375,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69344,
      "train_examples": 45000,
      "train_loss": 1.0471556855837505
    },
    {
      "epoch": 198,
      "eval_accuracy": 0.653,
      "eval_loss": 1.0403861328125,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69696,
      "train_examples": 45000,
      "train_loss": 1.0446702560000949
    },
    {
      "epoch": 199,
      "eval_accuracy": 0.6442,
      "eval_loss": 1.060380322265625,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70048,
      "train_examples": 45000,
      "train_loss": 1.0528953741285536
    },
    {
      "epoch": 200,
      "eval_accuracy": 0.6436,
      "eval_loss": 1.0704212158203126,
      "eval_split": "validation",
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70400,
      "train_examples": 45000,
      "train_loss": 1.0543489791870118
    }
  ],
  "evaluation_protocol": {
    "test_evaluations": 0,
    "test_used_for_selection": false,
    "validation_evaluations": 201
  },
  "final": {
    "accuracy": 0.6436,
    "epoch": 200,
    "evaluation_split": "validation",
    "finite": true,
    "loss": 1.0704212158203126,
    "step": 70400
  },
  "hardware": {
    "cuda_device_name": "NVIDIA RTX A6000",
    "cuda_visible_devices": "0",
    "device": "cuda",
    "device_total_memory_bytes": 50896568320,
    "peak_memory_allocated_bytes": 3252628992,
    "peak_memory_reserved_bytes": 4510973952,
    "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": 61.098384141922,
    "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": 3827.3347294330597,
    "train_wall_s": 3766.213252544403
  },
  "work": {
    "apical_macs_per_example": 125043328,
    "causal_scalar_observations": 0,
    "components": {
      "apical_projection_macs": 1125389952000000,
      "apical_regression_macs": 0,
      "bp_reverse_macs_estimate": 0,
      "kp_reciprocal_correlation_macs": 0,
      "local_weight_correlation_macs": 1129371264000000,
      "mirror_feedback_prediction_macs": 0,
      "mirror_local_correlation_macs": 0,
      "mirror_response_macs": 0,
      "ordinary_forward_macs": 1129371264000000,
      "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": 125485696,
    "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": 3384132480000000
  }
}