{
  "architecture": {
    "adaptive_apical_parameters": 2260992,
    "adaptive_feedback_parameters": 0,
    "base_width": 16,
    "blocks_per_stage": 3,
    "bn_eps": 1e-05,
    "bn_momentum": 0.1,
    "depth": 20,
    "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
    "fixed_feedback_parameters": 0,
    "fixed_traffic_coefficients": 188416,
    "forward_parameters": 269722,
    "hidden_shapes": [
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ]
    ],
    "normalization": "batchnorm",
    "predictor_parameters": 376832,
    "residual_scale": 1.0,
    "vectorizer_mode": "spatial_template",
    "vectorizer_parameters": 1884160
  },
  "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": 20,
    "device": "cuda",
    "epochs": 200,
    "eta_A": 0.01,
    "eta_P": 0.01,
    "eval_every": 1,
    "eval_split": "validation",
    "learn_P": 0,
    "loader_seed": 0,
    "lr": 0.01,
    "lr_gamma": 0.1,
    "lr_milestones": "100,150",
    "lr_schedule": "step",
    "max_steps": 0,
    "mirror_batch_size": 1,
    "mirror_eta": 0.1,
    "mirror_every": 16,
    "mirror_noise_std": 1.0,
    "mirror_seed": 3000,
    "mirror_warmup_steps": 0,
    "mode": "dfa",
    "momentum": 0.9,
    "neutral_projection": 0,
    "normalization": "batchnorm",
    "nuisance_scale": 0.0,
    "out": "/orion/u/oscarwan/scratch/sdil-work/sdil-a6000-full-rerun/results/resnet_crossover/r2/resnet-r2-dfa-d20-lr0p01.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": "full_hidden_field",
  "counters": {
    "apical_warmup_examples": 0,
    "calibration_event_examples": 0,
    "causal_scalar_observations": 0,
    "logical_batch_loss_queries": 0,
    "mirror_conv_examples": 0,
    "mirror_events": 0,
    "mirror_readout_examples": 0,
    "neutral_projection_examples": 0,
    "ordinary_examples": 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": -4.417266851911942e-05,
    "innovation_negative_gradient_cosine": [
      -0.00042737211333587766,
      0.0018140280153602362,
      0.0009411051869392395,
      -0.002272804267704487,
      -0.001042853808030486,
      0.0007228609756566584,
      -0.0014636360574513674,
      0.004689148627221584,
      0.00022076242021284997,
      0.0029502036049962044,
      0.006819497793912888,
      0.005044919438660145,
      0.007197106722742319,
      0.019712356850504875,
      0.021992990747094154,
      0.02051355317234993,
      0.026908088475465775,
      0.029072165489196777,
      0.039473552256822586
    ],
    "normalization_state": "training_batch_stats_without_running_update",
    "raw_negative_gradient_cosine": [
      -0.00042737211333587766,
      0.0018140280153602362,
      0.0009411051869392395,
      -0.002272804267704487,
      -0.001042853808030486,
      0.0007228609756566584,
      -0.0014636360574513674,
      0.004689148627221584,
      0.00022076242021284997,
      0.0029502036049962044,
      0.006819497793912888,
      0.005044919438660145,
      0.007197106722742319,
      0.019712356850504875,
      0.021992990747094154,
      0.02051355317234993,
      0.026908088475465775,
      0.029072165489196777,
      0.039473552256822586
    ],
    "teaching_negative_gradient_cosine": [
      -0.00042737211333587766,
      0.0018140280153602362,
      0.0009411051869392395,
      -0.002272804267704487,
      -0.001042853808030486,
      0.0007228609756566584,
      -0.0014636360574513674,
      0.004689148627221584,
      0.00022076242021284997,
      0.0029502036049962044,
      0.006819497793912888,
      0.005044919438660145,
      0.007197106722742319,
      0.019712356850504875,
      0.021992990747094154,
      0.02051355317234993,
      0.026908088475465775,
      0.029072165489196777,
      0.039473552256822586
    ],
    "wall_s": 0.13636183738708496
  },
  "epochs": [
    {
      "epoch": 1,
      "eval_accuracy": 0.2384,
      "eval_loss": 2.13388974609375,
      "eval_split": "validation",
      "lr": 0.002,
      "output_lr": 0.02,
      "step": 352,
      "train_examples": 45000,
      "train_loss": 2.196710070376926
    },
    {
      "epoch": 2,
      "eval_accuracy": 0.219,
      "eval_loss": 2.118686376953125,
      "eval_split": "validation",
      "lr": 0.004,
      "output_lr": 0.04,
      "step": 704,
      "train_examples": 45000,
      "train_loss": 2.1524469988505044
    },
    {
      "epoch": 3,
      "eval_accuracy": 0.2358,
      "eval_loss": 2.120386865234375,
      "eval_split": "validation",
      "lr": 0.006,
      "output_lr": 0.06000000000000001,
      "step": 1056,
      "train_examples": 45000,
      "train_loss": 2.16505956264072
    },
    {
      "epoch": 4,
      "eval_accuracy": 0.2538,
      "eval_loss": 2.0546119140625,
      "eval_split": "validation",
      "lr": 0.008,
      "output_lr": 0.08,
      "step": 1408,
      "train_examples": 45000,
      "train_loss": 2.250287850613064
    },
    {
      "epoch": 5,
      "eval_accuracy": 0.2272,
      "eval_loss": 2.4211458984375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 1760,
      "train_examples": 45000,
      "train_loss": 2.4122211878458657
    },
    {
      "epoch": 6,
      "eval_accuracy": 0.2418,
      "eval_loss": 2.48471376953125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 2112,
      "train_examples": 45000,
      "train_loss": 2.5066193411933053
    },
    {
      "epoch": 7,
      "eval_accuracy": 0.2642,
      "eval_loss": 2.71391015625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 2464,
      "train_examples": 45000,
      "train_loss": 3.1259462588416205
    },
    {
      "epoch": 8,
      "eval_accuracy": 0.2838,
      "eval_loss": 3.04355810546875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 2816,
      "train_examples": 45000,
      "train_loss": 3.7678212508307563
    },
    {
      "epoch": 9,
      "eval_accuracy": 0.2724,
      "eval_loss": 3.169311083984375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 3168,
      "train_examples": 45000,
      "train_loss": 3.511760868496365
    },
    {
      "epoch": 10,
      "eval_accuracy": 0.2212,
      "eval_loss": 7.113378125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 3520,
      "train_examples": 45000,
      "train_loss": 4.115778273179796
    },
    {
      "epoch": 11,
      "eval_accuracy": 0.2276,
      "eval_loss": 3.861981689453125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 3872,
      "train_examples": 45000,
      "train_loss": 5.214946193271213
    },
    {
      "epoch": 12,
      "eval_accuracy": 0.3098,
      "eval_loss": 3.261718896484375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 4224,
      "train_examples": 45000,
      "train_loss": 5.978883670467801
    },
    {
      "epoch": 13,
      "eval_accuracy": 0.2664,
      "eval_loss": 4.601353125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 4576,
      "train_examples": 45000,
      "train_loss": 6.004014733632405
    },
    {
      "epoch": 14,
      "eval_accuracy": 0.2076,
      "eval_loss": 8.8213029296875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 4928,
      "train_examples": 45000,
      "train_loss": 6.477945835537381
    },
    {
      "epoch": 15,
      "eval_accuracy": 0.1582,
      "eval_loss": 10.4473234375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 5280,
      "train_examples": 45000,
      "train_loss": 8.101313818952772
    },
    {
      "epoch": 16,
      "eval_accuracy": 0.3258,
      "eval_loss": 4.61436240234375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 5632,
      "train_examples": 45000,
      "train_loss": 8.165552450985379
    },
    {
      "epoch": 17,
      "eval_accuracy": 0.3088,
      "eval_loss": 10.7543853515625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 5984,
      "train_examples": 45000,
      "train_loss": 9.495957034132216
    },
    {
      "epoch": 18,
      "eval_accuracy": 0.2616,
      "eval_loss": 12.048984375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 6336,
      "train_examples": 45000,
      "train_loss": 10.429058261617024
    },
    {
      "epoch": 19,
      "eval_accuracy": 0.2384,
      "eval_loss": 11.6567517578125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 6688,
      "train_examples": 45000,
      "train_loss": 10.979766198221842
    },
    {
      "epoch": 20,
      "eval_accuracy": 0.2934,
      "eval_loss": 14.16575390625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 7040,
      "train_examples": 45000,
      "train_loss": 12.528717605929904
    },
    {
      "epoch": 21,
      "eval_accuracy": 0.232,
      "eval_loss": 26.35266015625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 7392,
      "train_examples": 45000,
      "train_loss": 14.306350407918295
    },
    {
      "epoch": 22,
      "eval_accuracy": 0.2194,
      "eval_loss": 22.857603125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 7744,
      "train_examples": 45000,
      "train_loss": 14.619046181911893
    },
    {
      "epoch": 23,
      "eval_accuracy": 0.2256,
      "eval_loss": 11.67859609375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 8096,
      "train_examples": 45000,
      "train_loss": 16.98607453087701
    },
    {
      "epoch": 24,
      "eval_accuracy": 0.2354,
      "eval_loss": 17.170334375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 8448,
      "train_examples": 45000,
      "train_loss": 19.04719298078749
    },
    {
      "epoch": 25,
      "eval_accuracy": 0.2598,
      "eval_loss": 15.6980568359375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 8800,
      "train_examples": 45000,
      "train_loss": 20.507712957763673
    },
    {
      "epoch": 26,
      "eval_accuracy": 0.2124,
      "eval_loss": 24.550714453125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 9152,
      "train_examples": 45000,
      "train_loss": 20.977451344129776
    },
    {
      "epoch": 27,
      "eval_accuracy": 0.2686,
      "eval_loss": 34.49674140625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 9504,
      "train_examples": 45000,
      "train_loss": 22.611929234483508
    },
    {
      "epoch": 28,
      "eval_accuracy": 0.2596,
      "eval_loss": 18.92392265625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 9856,
      "train_examples": 45000,
      "train_loss": 22.959516687011718
    },
    {
      "epoch": 29,
      "eval_accuracy": 0.2792,
      "eval_loss": 27.662224609375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 10208,
      "train_examples": 45000,
      "train_loss": 24.995158879937065
    },
    {
      "epoch": 30,
      "eval_accuracy": 0.2662,
      "eval_loss": 33.141328515625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 10560,
      "train_examples": 45000,
      "train_loss": 29.480212257554797
    },
    {
      "epoch": 31,
      "eval_accuracy": 0.226,
      "eval_loss": 27.803342578125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 10912,
      "train_examples": 45000,
      "train_loss": 26.761602039252388
    },
    {
      "epoch": 32,
      "eval_accuracy": 0.277,
      "eval_loss": 20.175499609375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 11264,
      "train_examples": 45000,
      "train_loss": 30.244316629367404
    },
    {
      "epoch": 33,
      "eval_accuracy": 0.3044,
      "eval_loss": 29.908916796875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 11616,
      "train_examples": 45000,
      "train_loss": 31.322398626708985
    },
    {
      "epoch": 34,
      "eval_accuracy": 0.245,
      "eval_loss": 45.0069875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 11968,
      "train_examples": 45000,
      "train_loss": 33.19071531711155
    },
    {
      "epoch": 35,
      "eval_accuracy": 0.3816,
      "eval_loss": 12.8741255859375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 12320,
      "train_examples": 45000,
      "train_loss": 35.12927552693685
    },
    {
      "epoch": 36,
      "eval_accuracy": 0.2418,
      "eval_loss": 62.24336953125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 12672,
      "train_examples": 45000,
      "train_loss": 38.10827041354709
    },
    {
      "epoch": 37,
      "eval_accuracy": 0.295,
      "eval_loss": 36.87111953125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 13024,
      "train_examples": 45000,
      "train_loss": 42.39294600694444
    },
    {
      "epoch": 38,
      "eval_accuracy": 0.1248,
      "eval_loss": 83.8959890625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 13376,
      "train_examples": 45000,
      "train_loss": 42.81453029039171
    },
    {
      "epoch": 39,
      "eval_accuracy": 0.289,
      "eval_loss": 85.4705140625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 13728,
      "train_examples": 45000,
      "train_loss": 46.396105011664496
    },
    {
      "epoch": 40,
      "eval_accuracy": 0.224,
      "eval_loss": 59.92510703125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 14080,
      "train_examples": 45000,
      "train_loss": 46.26691556532118
    },
    {
      "epoch": 41,
      "eval_accuracy": 0.2546,
      "eval_loss": 37.7355953125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 14432,
      "train_examples": 45000,
      "train_loss": 50.41342158237033
    },
    {
      "epoch": 42,
      "eval_accuracy": 0.2456,
      "eval_loss": 77.8837359375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 14784,
      "train_examples": 45000,
      "train_loss": 50.42762740614149
    },
    {
      "epoch": 43,
      "eval_accuracy": 0.2972,
      "eval_loss": 37.9858015625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 15136,
      "train_examples": 45000,
      "train_loss": 55.54701350640191
    },
    {
      "epoch": 44,
      "eval_accuracy": 0.3668,
      "eval_loss": 43.6133109375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 15488,
      "train_examples": 45000,
      "train_loss": 62.59512099134657
    },
    {
      "epoch": 45,
      "eval_accuracy": 0.2564,
      "eval_loss": 53.872640625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 15840,
      "train_examples": 45000,
      "train_loss": 56.61767745632596
    },
    {
      "epoch": 46,
      "eval_accuracy": 0.1912,
      "eval_loss": 132.7386125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 16192,
      "train_examples": 45000,
      "train_loss": 70.22152931179471
    },
    {
      "epoch": 47,
      "eval_accuracy": 0.368,
      "eval_loss": 35.1759109375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 16544,
      "train_examples": 45000,
      "train_loss": 70.48875649888781
    },
    {
      "epoch": 48,
      "eval_accuracy": 0.365,
      "eval_loss": 30.53370625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 16896,
      "train_examples": 45000,
      "train_loss": 69.93644878607856
    },
    {
      "epoch": 49,
      "eval_accuracy": 0.283,
      "eval_loss": 57.0330375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 17248,
      "train_examples": 45000,
      "train_loss": 77.78397584296333
    },
    {
      "epoch": 50,
      "eval_accuracy": 0.2704,
      "eval_loss": 118.6991953125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 17600,
      "train_examples": 45000,
      "train_loss": 75.2017873196072
    },
    {
      "epoch": 51,
      "eval_accuracy": 0.316,
      "eval_loss": 63.6082796875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 17952,
      "train_examples": 45000,
      "train_loss": 83.56330694444445
    },
    {
      "epoch": 52,
      "eval_accuracy": 0.2914,
      "eval_loss": 108.32306875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 18304,
      "train_examples": 45000,
      "train_loss": 86.26330736897786
    },
    {
      "epoch": 53,
      "eval_accuracy": 0.334,
      "eval_loss": 46.92616015625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 18656,
      "train_examples": 45000,
      "train_loss": 87.97140079684787
    },
    {
      "epoch": 54,
      "eval_accuracy": 0.2518,
      "eval_loss": 104.82296875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 19008,
      "train_examples": 45000,
      "train_loss": 93.9687024102105
    },
    {
      "epoch": 55,
      "eval_accuracy": 0.3186,
      "eval_loss": 61.5724078125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 19360,
      "train_examples": 45000,
      "train_loss": 96.0325408203125
    },
    {
      "epoch": 56,
      "eval_accuracy": 0.3622,
      "eval_loss": 54.3312796875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 19712,
      "train_examples": 45000,
      "train_loss": 103.135226820204
    },
    {
      "epoch": 57,
      "eval_accuracy": 0.2204,
      "eval_loss": 214.8653625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 20064,
      "train_examples": 45000,
      "train_loss": 112.44556560058594
    },
    {
      "epoch": 58,
      "eval_accuracy": 0.336,
      "eval_loss": 76.492653125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 20416,
      "train_examples": 45000,
      "train_loss": 97.48807627766926
    },
    {
      "epoch": 59,
      "eval_accuracy": 0.295,
      "eval_loss": 134.529453125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 20768,
      "train_examples": 45000,
      "train_loss": 115.6962455078125
    },
    {
      "epoch": 60,
      "eval_accuracy": 0.3338,
      "eval_loss": 80.4154484375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 21120,
      "train_examples": 45000,
      "train_loss": 114.1658334906684
    },
    {
      "epoch": 61,
      "eval_accuracy": 0.2612,
      "eval_loss": 82.935240625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 21472,
      "train_examples": 45000,
      "train_loss": 120.49449004584419
    },
    {
      "epoch": 62,
      "eval_accuracy": 0.2506,
      "eval_loss": 177.4444,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 21824,
      "train_examples": 45000,
      "train_loss": 129.52251632215712
    },
    {
      "epoch": 63,
      "eval_accuracy": 0.2992,
      "eval_loss": 158.935075,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 22176,
      "train_examples": 45000,
      "train_loss": 126.56348369276259
    },
    {
      "epoch": 64,
      "eval_accuracy": 0.2318,
      "eval_loss": 224.766190625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 22528,
      "train_examples": 45000,
      "train_loss": 129.15274677734374
    },
    {
      "epoch": 65,
      "eval_accuracy": 0.258,
      "eval_loss": 149.426728125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 22880,
      "train_examples": 45000,
      "train_loss": 149.42977838812934
    },
    {
      "epoch": 66,
      "eval_accuracy": 0.1468,
      "eval_loss": 236.724253125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 23232,
      "train_examples": 45000,
      "train_loss": 137.12026451009115
    },
    {
      "epoch": 67,
      "eval_accuracy": 0.288,
      "eval_loss": 133.139496875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 23584,
      "train_examples": 45000,
      "train_loss": 164.30003099907768
    },
    {
      "epoch": 68,
      "eval_accuracy": 0.2536,
      "eval_loss": 175.942903125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 23936,
      "train_examples": 45000,
      "train_loss": 146.02838853624132
    },
    {
      "epoch": 69,
      "eval_accuracy": 0.2536,
      "eval_loss": 117.812740625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 24288,
      "train_examples": 45000,
      "train_loss": 170.32516332465278
    },
    {
      "epoch": 70,
      "eval_accuracy": 0.3408,
      "eval_loss": 139.700415625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 24640,
      "train_examples": 45000,
      "train_loss": 167.26187187771268
    },
    {
      "epoch": 71,
      "eval_accuracy": 0.275,
      "eval_loss": 271.8581,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 24992,
      "train_examples": 45000,
      "train_loss": 150.94737664930557
    },
    {
      "epoch": 72,
      "eval_accuracy": 0.2768,
      "eval_loss": 109.6359796875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 25344,
      "train_examples": 45000,
      "train_loss": 178.7975426405165
    },
    {
      "epoch": 73,
      "eval_accuracy": 0.2492,
      "eval_loss": 153.839534375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 25696,
      "train_examples": 45000,
      "train_loss": 181.78071877712674
    },
    {
      "epoch": 74,
      "eval_accuracy": 0.2398,
      "eval_loss": 362.14533125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 26048,
      "train_examples": 45000,
      "train_loss": 179.54143683810764
    },
    {
      "epoch": 75,
      "eval_accuracy": 0.333,
      "eval_loss": 109.10584375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 26400,
      "train_examples": 45000,
      "train_loss": 228.60323557128908
    },
    {
      "epoch": 76,
      "eval_accuracy": 0.1564,
      "eval_loss": 465.6028625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 26752,
      "train_examples": 45000,
      "train_loss": 178.3261564453125
    },
    {
      "epoch": 77,
      "eval_accuracy": 0.3438,
      "eval_loss": 100.3464203125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 27104,
      "train_examples": 45000,
      "train_loss": 208.42907971191406
    },
    {
      "epoch": 78,
      "eval_accuracy": 0.2746,
      "eval_loss": 294.67795,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 27456,
      "train_examples": 45000,
      "train_loss": 210.42169096137152
    },
    {
      "epoch": 79,
      "eval_accuracy": 0.276,
      "eval_loss": 190.441990625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 27808,
      "train_examples": 45000,
      "train_loss": 224.93065251193576
    },
    {
      "epoch": 80,
      "eval_accuracy": 0.3246,
      "eval_loss": 248.689990625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 28160,
      "train_examples": 45000,
      "train_loss": 222.3671440999349
    },
    {
      "epoch": 81,
      "eval_accuracy": 0.271,
      "eval_loss": 167.538346875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 28512,
      "train_examples": 45000,
      "train_loss": 235.71425759006075
    },
    {
      "epoch": 82,
      "eval_accuracy": 0.1452,
      "eval_loss": 355.31036875,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 28864,
      "train_examples": 45000,
      "train_loss": 222.38550850423178
    },
    {
      "epoch": 83,
      "eval_accuracy": 0.2744,
      "eval_loss": 190.79715625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 29216,
      "train_examples": 45000,
      "train_loss": 234.47545372721353
    },
    {
      "epoch": 84,
      "eval_accuracy": 0.2446,
      "eval_loss": 268.720678125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 29568,
      "train_examples": 45000,
      "train_loss": 239.26784101833766
    },
    {
      "epoch": 85,
      "eval_accuracy": 0.2664,
      "eval_loss": 219.119975,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 29920,
      "train_examples": 45000,
      "train_loss": 277.5253189887153
    },
    {
      "epoch": 86,
      "eval_accuracy": 0.2812,
      "eval_loss": 186.918509375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 30272,
      "train_examples": 45000,
      "train_loss": 267.0558938340929
    },
    {
      "epoch": 87,
      "eval_accuracy": 0.275,
      "eval_loss": 291.21615625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 30624,
      "train_examples": 45000,
      "train_loss": 253.1571425998264
    },
    {
      "epoch": 88,
      "eval_accuracy": 0.4022,
      "eval_loss": 197.7845375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 30976,
      "train_examples": 45000,
      "train_loss": 271.55762882486977
    },
    {
      "epoch": 89,
      "eval_accuracy": 0.2136,
      "eval_loss": 384.08820625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 31328,
      "train_examples": 45000,
      "train_loss": 272.41770344509547
    },
    {
      "epoch": 90,
      "eval_accuracy": 0.2894,
      "eval_loss": 266.3597,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 31680,
      "train_examples": 45000,
      "train_loss": 279.4282325954861
    },
    {
      "epoch": 91,
      "eval_accuracy": 0.3094,
      "eval_loss": 177.42945,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 32032,
      "train_examples": 45000,
      "train_loss": 307.67940743001304
    },
    {
      "epoch": 92,
      "eval_accuracy": 0.1138,
      "eval_loss": 565.199775,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 32384,
      "train_examples": 45000,
      "train_loss": 308.0857044731988
    },
    {
      "epoch": 93,
      "eval_accuracy": 0.2484,
      "eval_loss": 306.35499375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 32736,
      "train_examples": 45000,
      "train_loss": 296.103281765408
    },
    {
      "epoch": 94,
      "eval_accuracy": 0.3628,
      "eval_loss": 315.681325,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 33088,
      "train_examples": 45000,
      "train_loss": 322.4415712673611
    },
    {
      "epoch": 95,
      "eval_accuracy": 0.3176,
      "eval_loss": 212.66410625,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 33440,
      "train_examples": 45000,
      "train_loss": 327.5514536295573
    },
    {
      "epoch": 96,
      "eval_accuracy": 0.345,
      "eval_loss": 154.168359375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 33792,
      "train_examples": 45000,
      "train_loss": 375.67241025661895
    },
    {
      "epoch": 97,
      "eval_accuracy": 0.364,
      "eval_loss": 292.67804375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 34144,
      "train_examples": 45000,
      "train_loss": 345.29349494900174
    },
    {
      "epoch": 98,
      "eval_accuracy": 0.3088,
      "eval_loss": 412.27108125,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 34496,
      "train_examples": 45000,
      "train_loss": 366.83919205729165
    },
    {
      "epoch": 99,
      "eval_accuracy": 0.2158,
      "eval_loss": 538.530225,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 34848,
      "train_examples": 45000,
      "train_loss": 346.40431893988716
    },
    {
      "epoch": 100,
      "eval_accuracy": 0.2714,
      "eval_loss": 350.2196375,
      "eval_split": "validation",
      "lr": 0.01,
      "output_lr": 0.1,
      "step": 35200,
      "train_examples": 45000,
      "train_loss": 364.6515966362847
    },
    {
      "epoch": 101,
      "eval_accuracy": 0.4338,
      "eval_loss": 72.9980640625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 35552,
      "train_examples": 45000,
      "train_loss": 97.5824759250217
    },
    {
      "epoch": 102,
      "eval_accuracy": 0.3946,
      "eval_loss": 63.793271875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 35904,
      "train_examples": 45000,
      "train_loss": 68.66056476982965
    },
    {
      "epoch": 103,
      "eval_accuracy": 0.3972,
      "eval_loss": 54.701396875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 36256,
      "train_examples": 45000,
      "train_loss": 58.456144613308375
    },
    {
      "epoch": 104,
      "eval_accuracy": 0.395,
      "eval_loss": 45.80713984375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 36608,
      "train_examples": 45000,
      "train_loss": 52.66843123033312
    },
    {
      "epoch": 105,
      "eval_accuracy": 0.3904,
      "eval_loss": 42.960628125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 36960,
      "train_examples": 45000,
      "train_loss": 47.554370183648004
    },
    {
      "epoch": 106,
      "eval_accuracy": 0.3928,
      "eval_loss": 37.1106109375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 37312,
      "train_examples": 45000,
      "train_loss": 43.874717364501954
    },
    {
      "epoch": 107,
      "eval_accuracy": 0.4084,
      "eval_loss": 33.58920625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 37664,
      "train_examples": 45000,
      "train_loss": 42.896230186631946
    },
    {
      "epoch": 108,
      "eval_accuracy": 0.3816,
      "eval_loss": 35.4541578125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 38016,
      "train_examples": 45000,
      "train_loss": 40.49658525119358
    },
    {
      "epoch": 109,
      "eval_accuracy": 0.3904,
      "eval_loss": 31.881768359375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 38368,
      "train_examples": 45000,
      "train_loss": 38.327581633843316
    },
    {
      "epoch": 110,
      "eval_accuracy": 0.3458,
      "eval_loss": 38.830978125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 38720,
      "train_examples": 45000,
      "train_loss": 35.389524947781034
    },
    {
      "epoch": 111,
      "eval_accuracy": 0.3544,
      "eval_loss": 36.43687421875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 39072,
      "train_examples": 45000,
      "train_loss": 34.1201142171224
    },
    {
      "epoch": 112,
      "eval_accuracy": 0.3744,
      "eval_loss": 27.77460859375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 39424,
      "train_examples": 45000,
      "train_loss": 32.8625382297092
    },
    {
      "epoch": 113,
      "eval_accuracy": 0.3666,
      "eval_loss": 34.01153984375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 39776,
      "train_examples": 45000,
      "train_loss": 32.72871079440647
    },
    {
      "epoch": 114,
      "eval_accuracy": 0.3624,
      "eval_loss": 29.06900859375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 40128,
      "train_examples": 45000,
      "train_loss": 31.676572363620334
    },
    {
      "epoch": 115,
      "eval_accuracy": 0.2994,
      "eval_loss": 38.16306328125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 40480,
      "train_examples": 45000,
      "train_loss": 30.014509815131294
    },
    {
      "epoch": 116,
      "eval_accuracy": 0.348,
      "eval_loss": 27.01443515625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 40832,
      "train_examples": 45000,
      "train_loss": 30.27423641289605
    },
    {
      "epoch": 117,
      "eval_accuracy": 0.3002,
      "eval_loss": 45.1408640625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 41184,
      "train_examples": 45000,
      "train_loss": 30.936224088541667
    },
    {
      "epoch": 118,
      "eval_accuracy": 0.3548,
      "eval_loss": 34.59181484375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 41536,
      "train_examples": 45000,
      "train_loss": 30.146907942708335
    },
    {
      "epoch": 119,
      "eval_accuracy": 0.3376,
      "eval_loss": 32.60268515625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 41888,
      "train_examples": 45000,
      "train_loss": 28.432915776231553
    },
    {
      "epoch": 120,
      "eval_accuracy": 0.3364,
      "eval_loss": 34.4127140625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 42240,
      "train_examples": 45000,
      "train_loss": 29.565025434705944
    },
    {
      "epoch": 121,
      "eval_accuracy": 0.3672,
      "eval_loss": 21.1856453125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 42592,
      "train_examples": 45000,
      "train_loss": 28.357878097195094
    },
    {
      "epoch": 122,
      "eval_accuracy": 0.3018,
      "eval_loss": 31.5717625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 42944,
      "train_examples": 45000,
      "train_loss": 29.79608405388726
    },
    {
      "epoch": 123,
      "eval_accuracy": 0.3136,
      "eval_loss": 28.0333640625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 43296,
      "train_examples": 45000,
      "train_loss": 32.39813755900065
    },
    {
      "epoch": 124,
      "eval_accuracy": 0.33,
      "eval_loss": 25.2027203125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 43648,
      "train_examples": 45000,
      "train_loss": 28.785149905225964
    },
    {
      "epoch": 125,
      "eval_accuracy": 0.365,
      "eval_loss": 50.01820859375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 44000,
      "train_examples": 45000,
      "train_loss": 30.066769323052302
    },
    {
      "epoch": 126,
      "eval_accuracy": 0.2992,
      "eval_loss": 38.751096875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 44352,
      "train_examples": 45000,
      "train_loss": 28.825669424777562
    },
    {
      "epoch": 127,
      "eval_accuracy": 0.3188,
      "eval_loss": 31.16794609375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 44704,
      "train_examples": 45000,
      "train_loss": 31.951569648573134
    },
    {
      "epoch": 128,
      "eval_accuracy": 0.3536,
      "eval_loss": 26.708534375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 45056,
      "train_examples": 45000,
      "train_loss": 28.16607522176107
    },
    {
      "epoch": 129,
      "eval_accuracy": 0.3712,
      "eval_loss": 22.56696171875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 45408,
      "train_examples": 45000,
      "train_loss": 30.68563767361111
    },
    {
      "epoch": 130,
      "eval_accuracy": 0.3722,
      "eval_loss": 27.211113671875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 45760,
      "train_examples": 45000,
      "train_loss": 26.697583384874132
    },
    {
      "epoch": 131,
      "eval_accuracy": 0.3952,
      "eval_loss": 15.98902734375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 46112,
      "train_examples": 45000,
      "train_loss": 28.169558378770617
    },
    {
      "epoch": 132,
      "eval_accuracy": 0.345,
      "eval_loss": 27.111884375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 46464,
      "train_examples": 45000,
      "train_loss": 29.73483961995443
    },
    {
      "epoch": 133,
      "eval_accuracy": 0.3092,
      "eval_loss": 33.95772265625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 46816,
      "train_examples": 45000,
      "train_loss": 35.36185829772949
    },
    {
      "epoch": 134,
      "eval_accuracy": 0.3584,
      "eval_loss": 20.575744921875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 47168,
      "train_examples": 45000,
      "train_loss": 30.647867134602865
    },
    {
      "epoch": 135,
      "eval_accuracy": 0.3062,
      "eval_loss": 25.4248328125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 47520,
      "train_examples": 45000,
      "train_loss": 26.337036939154732
    },
    {
      "epoch": 136,
      "eval_accuracy": 0.4092,
      "eval_loss": 15.70295703125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 47872,
      "train_examples": 45000,
      "train_loss": 33.879409608968096
    },
    {
      "epoch": 137,
      "eval_accuracy": 0.2672,
      "eval_loss": 40.63083671875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 48224,
      "train_examples": 45000,
      "train_loss": 30.210656193712023
    },
    {
      "epoch": 138,
      "eval_accuracy": 0.2778,
      "eval_loss": 34.45653125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 48576,
      "train_examples": 45000,
      "train_loss": 34.42727603759766
    },
    {
      "epoch": 139,
      "eval_accuracy": 0.3152,
      "eval_loss": 41.16488515625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 48928,
      "train_examples": 45000,
      "train_loss": 30.716559368896483
    },
    {
      "epoch": 140,
      "eval_accuracy": 0.293,
      "eval_loss": 29.28139296875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 49280,
      "train_examples": 45000,
      "train_loss": 30.197615693155925
    },
    {
      "epoch": 141,
      "eval_accuracy": 0.2916,
      "eval_loss": 32.38239140625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 49632,
      "train_examples": 45000,
      "train_loss": 28.30102506205241
    },
    {
      "epoch": 142,
      "eval_accuracy": 0.2882,
      "eval_loss": 31.02913203125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 49984,
      "train_examples": 45000,
      "train_loss": 33.08134716661241
    },
    {
      "epoch": 143,
      "eval_accuracy": 0.3546,
      "eval_loss": 26.13366640625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 50336,
      "train_examples": 45000,
      "train_loss": 35.202170224338104
    },
    {
      "epoch": 144,
      "eval_accuracy": 0.4018,
      "eval_loss": 18.689922265625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 50688,
      "train_examples": 45000,
      "train_loss": 28.16269857449002
    },
    {
      "epoch": 145,
      "eval_accuracy": 0.2002,
      "eval_loss": 45.4462328125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 51040,
      "train_examples": 45000,
      "train_loss": 29.617123545328777
    },
    {
      "epoch": 146,
      "eval_accuracy": 0.3582,
      "eval_loss": 24.901905078125,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 51392,
      "train_examples": 45000,
      "train_loss": 34.85614291212294
    },
    {
      "epoch": 147,
      "eval_accuracy": 0.3712,
      "eval_loss": 39.63477265625,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 51744,
      "train_examples": 45000,
      "train_loss": 32.07841902872721
    },
    {
      "epoch": 148,
      "eval_accuracy": 0.3844,
      "eval_loss": 23.235046875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 52096,
      "train_examples": 45000,
      "train_loss": 34.80584049445258
    },
    {
      "epoch": 149,
      "eval_accuracy": 0.3602,
      "eval_loss": 31.175896484375,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 52448,
      "train_examples": 45000,
      "train_loss": 32.27111835700141
    },
    {
      "epoch": 150,
      "eval_accuracy": 0.2914,
      "eval_loss": 28.874794921875,
      "eval_split": "validation",
      "lr": 0.001,
      "output_lr": 0.010000000000000002,
      "step": 52800,
      "train_examples": 45000,
      "train_loss": 31.977565850151908
    },
    {
      "epoch": 151,
      "eval_accuracy": 0.4476,
      "eval_loss": 10.8708392578125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 53152,
      "train_examples": 45000,
      "train_loss": 11.651757261149088
    },
    {
      "epoch": 152,
      "eval_accuracy": 0.446,
      "eval_loss": 9.7358521484375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 53504,
      "train_examples": 45000,
      "train_loss": 9.99678972998725
    },
    {
      "epoch": 153,
      "eval_accuracy": 0.4404,
      "eval_loss": 8.9593001953125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 53856,
      "train_examples": 45000,
      "train_loss": 9.319264670477972
    },
    {
      "epoch": 154,
      "eval_accuracy": 0.4428,
      "eval_loss": 8.37555732421875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 54208,
      "train_examples": 45000,
      "train_loss": 8.697566080050999
    },
    {
      "epoch": 155,
      "eval_accuracy": 0.4502,
      "eval_loss": 7.80713427734375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 54560,
      "train_examples": 45000,
      "train_loss": 8.319801394992405
    },
    {
      "epoch": 156,
      "eval_accuracy": 0.4344,
      "eval_loss": 7.308902734375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 54912,
      "train_examples": 45000,
      "train_loss": 7.8531677338494195
    },
    {
      "epoch": 157,
      "eval_accuracy": 0.4406,
      "eval_loss": 6.8099802734375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 55264,
      "train_examples": 45000,
      "train_loss": 7.516104418097602
    },
    {
      "epoch": 158,
      "eval_accuracy": 0.4278,
      "eval_loss": 6.6249916015625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 55616,
      "train_examples": 45000,
      "train_loss": 7.2278142467922635
    },
    {
      "epoch": 159,
      "eval_accuracy": 0.4346,
      "eval_loss": 6.33050771484375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 55968,
      "train_examples": 45000,
      "train_loss": 6.960628581576877
    },
    {
      "epoch": 160,
      "eval_accuracy": 0.4234,
      "eval_loss": 6.38564306640625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 56320,
      "train_examples": 45000,
      "train_loss": 6.7058130749172635
    },
    {
      "epoch": 161,
      "eval_accuracy": 0.4256,
      "eval_loss": 5.79991982421875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 56672,
      "train_examples": 45000,
      "train_loss": 6.504179000430637
    },
    {
      "epoch": 162,
      "eval_accuracy": 0.408,
      "eval_loss": 5.9061658203125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 57024,
      "train_examples": 45000,
      "train_loss": 6.2047558926052515
    },
    {
      "epoch": 163,
      "eval_accuracy": 0.4004,
      "eval_loss": 6.0522150390625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 57376,
      "train_examples": 45000,
      "train_loss": 6.164733709801568
    },
    {
      "epoch": 164,
      "eval_accuracy": 0.4006,
      "eval_loss": 5.61034873046875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 57728,
      "train_examples": 45000,
      "train_loss": 5.856545046403673
    },
    {
      "epoch": 165,
      "eval_accuracy": 0.3968,
      "eval_loss": 5.4717162109375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 58080,
      "train_examples": 45000,
      "train_loss": 5.771546636369493
    },
    {
      "epoch": 166,
      "eval_accuracy": 0.3862,
      "eval_loss": 5.516412109375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 58432,
      "train_examples": 45000,
      "train_loss": 5.6266596131218805
    },
    {
      "epoch": 167,
      "eval_accuracy": 0.4108,
      "eval_loss": 4.80401162109375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 58784,
      "train_examples": 45000,
      "train_loss": 5.476546478610569
    },
    {
      "epoch": 168,
      "eval_accuracy": 0.3934,
      "eval_loss": 5.0674400390625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59136,
      "train_examples": 45000,
      "train_loss": 5.386792821163601
    },
    {
      "epoch": 169,
      "eval_accuracy": 0.3716,
      "eval_loss": 5.19190400390625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59488,
      "train_examples": 45000,
      "train_loss": 5.217026385074192
    },
    {
      "epoch": 170,
      "eval_accuracy": 0.4034,
      "eval_loss": 4.618946875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59840,
      "train_examples": 45000,
      "train_loss": 5.097870582156711
    },
    {
      "epoch": 171,
      "eval_accuracy": 0.3978,
      "eval_loss": 4.6441703125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60192,
      "train_examples": 45000,
      "train_loss": 4.992071345859104
    },
    {
      "epoch": 172,
      "eval_accuracy": 0.415,
      "eval_loss": 4.309574365234375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60544,
      "train_examples": 45000,
      "train_loss": 5.006023575761583
    },
    {
      "epoch": 173,
      "eval_accuracy": 0.3728,
      "eval_loss": 4.7692791015625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60896,
      "train_examples": 45000,
      "train_loss": 4.765749631924099
    },
    {
      "epoch": 174,
      "eval_accuracy": 0.3972,
      "eval_loss": 4.273984814453125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61248,
      "train_examples": 45000,
      "train_loss": 4.669230929056804
    },
    {
      "epoch": 175,
      "eval_accuracy": 0.3868,
      "eval_loss": 4.346198095703125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61600,
      "train_examples": 45000,
      "train_loss": 4.767588217163086
    },
    {
      "epoch": 176,
      "eval_accuracy": 0.4138,
      "eval_loss": 4.04446513671875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61952,
      "train_examples": 45000,
      "train_loss": 4.574014286422729
    },
    {
      "epoch": 177,
      "eval_accuracy": 0.3986,
      "eval_loss": 4.01624287109375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62304,
      "train_examples": 45000,
      "train_loss": 4.483139426252577
    },
    {
      "epoch": 178,
      "eval_accuracy": 0.379,
      "eval_loss": 4.116311572265625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62656,
      "train_examples": 45000,
      "train_loss": 4.440480245547825
    },
    {
      "epoch": 179,
      "eval_accuracy": 0.411,
      "eval_loss": 3.7937888671875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63008,
      "train_examples": 45000,
      "train_loss": 4.464287152438693
    },
    {
      "epoch": 180,
      "eval_accuracy": 0.3336,
      "eval_loss": 4.47425,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63360,
      "train_examples": 45000,
      "train_loss": 4.374478776211209
    },
    {
      "epoch": 181,
      "eval_accuracy": 0.3926,
      "eval_loss": 3.835095654296875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63712,
      "train_examples": 45000,
      "train_loss": 4.33659748111301
    },
    {
      "epoch": 182,
      "eval_accuracy": 0.3928,
      "eval_loss": 3.6995392578125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64064,
      "train_examples": 45000,
      "train_loss": 4.252237670898437
    },
    {
      "epoch": 183,
      "eval_accuracy": 0.3914,
      "eval_loss": 3.63160205078125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64416,
      "train_examples": 45000,
      "train_loss": 4.279515455034044
    },
    {
      "epoch": 184,
      "eval_accuracy": 0.3968,
      "eval_loss": 3.59172255859375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64768,
      "train_examples": 45000,
      "train_loss": 4.1540915985107425
    },
    {
      "epoch": 185,
      "eval_accuracy": 0.3818,
      "eval_loss": 3.789047998046875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65120,
      "train_examples": 45000,
      "train_loss": 4.0526812603420685
    },
    {
      "epoch": 186,
      "eval_accuracy": 0.361,
      "eval_loss": 4.066609326171875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65472,
      "train_examples": 45000,
      "train_loss": 4.036473704528809
    },
    {
      "epoch": 187,
      "eval_accuracy": 0.3994,
      "eval_loss": 3.52081162109375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65824,
      "train_examples": 45000,
      "train_loss": 4.109895914713542
    },
    {
      "epoch": 188,
      "eval_accuracy": 0.3908,
      "eval_loss": 3.533601025390625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66176,
      "train_examples": 45000,
      "train_loss": 4.0574144656711155
    },
    {
      "epoch": 189,
      "eval_accuracy": 0.3962,
      "eval_loss": 3.42502001953125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66528,
      "train_examples": 45000,
      "train_loss": 4.032029030354818
    },
    {
      "epoch": 190,
      "eval_accuracy": 0.4156,
      "eval_loss": 3.239801220703125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66880,
      "train_examples": 45000,
      "train_loss": 3.874290327750312
    },
    {
      "epoch": 191,
      "eval_accuracy": 0.3698,
      "eval_loss": 3.478581982421875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67232,
      "train_examples": 45000,
      "train_loss": 3.7694188574896916
    },
    {
      "epoch": 192,
      "eval_accuracy": 0.3578,
      "eval_loss": 3.897442578125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67584,
      "train_examples": 45000,
      "train_loss": 3.7616168645222983
    },
    {
      "epoch": 193,
      "eval_accuracy": 0.3774,
      "eval_loss": 3.65839599609375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67936,
      "train_examples": 45000,
      "train_loss": 3.8300268634372285
    },
    {
      "epoch": 194,
      "eval_accuracy": 0.3948,
      "eval_loss": 3.29561572265625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68288,
      "train_examples": 45000,
      "train_loss": 3.8239806078592937
    },
    {
      "epoch": 195,
      "eval_accuracy": 0.3592,
      "eval_loss": 3.819576904296875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68640,
      "train_examples": 45000,
      "train_loss": 3.7138434155782063
    },
    {
      "epoch": 196,
      "eval_accuracy": 0.4016,
      "eval_loss": 3.35995888671875,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68992,
      "train_examples": 45000,
      "train_loss": 3.6911021950191922
    },
    {
      "epoch": 197,
      "eval_accuracy": 0.3778,
      "eval_loss": 3.279795556640625,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69344,
      "train_examples": 45000,
      "train_loss": 3.6857616521199543
    },
    {
      "epoch": 198,
      "eval_accuracy": 0.3356,
      "eval_loss": 4.24252001953125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69696,
      "train_examples": 45000,
      "train_loss": 3.709500290086534
    },
    {
      "epoch": 199,
      "eval_accuracy": 0.3088,
      "eval_loss": 3.90576396484375,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70048,
      "train_examples": 45000,
      "train_loss": 3.7073160080379908
    },
    {
      "epoch": 200,
      "eval_accuracy": 0.3492,
      "eval_loss": 3.6838392578125,
      "eval_split": "validation",
      "lr": 0.00010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70400,
      "train_examples": 45000,
      "train_loss": 3.5896300240410697
    }
  ],
  "evaluation_protocol": {
    "test_evaluations": 0,
    "test_used_for_selection": false,
    "validation_evaluations": 201
  },
  "final": {
    "accuracy": 0.3492,
    "epoch": 200,
    "evaluation_split": "validation",
    "finite": true,
    "loss": 3.6838392578125,
    "step": 70400
  },
  "hardware": {
    "cuda_device_name": "NVIDIA RTX A6000",
    "cuda_visible_devices": "0",
    "device": "cuda",
    "device_total_memory_bytes": 50896568320,
    "peak_memory_allocated_bytes": 2018397696,
    "peak_memory_reserved_bytes": 2480930816,
    "torch_version": "2.3.1+cu118"
  },
  "protocol_family": "oral_a_cifar_local_resnet_development",
  "provenance": {
    "git_commit": "f55872beaee658ff76ac15fd421b73017820c320",
    "git_tracked_dirty": false
  },
  "schema_version": 1,
  "split": {
    "cifar_source_files": [
      {
        "bytes": 31035704,
        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_1",
        "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
      },
      {
        "bytes": 31035320,
        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_2",
        "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
      },
      {
        "bytes": 31035999,
        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_3",
        "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
      },
      {
        "bytes": 31035696,
        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_4",
        "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
      },
      {
        "bytes": 31035623,
        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_5",
        "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
      },
      {
        "bytes": 31035526,
        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/test_batch",
        "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831"
      }
    ],
    "dataset": "cifar10",
    "evaluation_split": "validation",
    "input_layout": "NCHW",
    "input_shape": [
      3,
      32,
      32
    ],
    "loader_seed": 0,
    "normalization_mean": [
      0.49140000343322754,
      0.4821999967098236,
      0.4465000033378601
    ],
    "normalization_std": [
      0.24699999392032623,
      0.2434999942779541,
      0.26159998774528503
    ],
    "split_from_training_only": true,
    "split_seed": 2027,
    "test_examples": 10000,
    "train_examples": 45000,
    "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
    "validation_class_counts": {
      "0": 500,
      "1": 500,
      "2": 500,
      "3": 500,
      "4": 500,
      "5": 500,
      "6": 500,
      "7": 500,
      "8": 500,
      "9": 500
    },
    "validation_examples": 5000,
    "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b"
  },
  "timing": {
    "apical_warmup_wall_s": 0.0,
    "evaluation_wall_s": 21.157899856567383,
    "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": 1183.4196124076843,
    "train_wall_s": 1162.2410652637482
  },
  "work": {
    "apical_macs_per_example": 1884160,
    "causal_scalar_observations": 0,
    "components": {
      "apical_projection_macs": 16957440000000,
      "apical_regression_macs": 0,
      "bp_reverse_macs_estimate": 0,
      "kp_reciprocal_correlation_macs": 0,
      "local_weight_correlation_macs": 364959360000000,
      "mirror_feedback_prediction_macs": 0,
      "mirror_local_correlation_macs": 0,
      "mirror_response_macs": 0,
      "ordinary_forward_macs": 364959360000000,
      "perturbation_forward_macs": 0,
      "warmup_clean_forward_macs": 0
    },
    "definition": "multiply-accumulates in conv/linear maps; one local weight correlation equals one forward-weight MAC count; BP reverse is estimated as one weight-gradient plus one activation-gradient convolution per forward convolution; mixed-traffic/predictor elementwise arithmetic is reported as a conservative operation estimate separately and is not folded into MACs",
    "elementwise_operations_estimate": 0,
    "forward_macs_per_example": 40551040,
    "logical_batch_loss_queries": 0,
    "mirror_probe_examples": 0,
    "mirror_readout_probe_examples": 0,
    "neutral_projection_observations": 0,
    "per_example_cross_entropy_terms": 0,
    "total_clean_forward_examples": 9000000,
    "total_forward_equivalent_examples": 9000000,
    "total_macs_estimate": 746876160000000
  }
}