{
  "architecture": {
    "adaptive_apical_parameters": 6389760,
    "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": 0,
    "fixed_traffic_coefficients": 532480,
    "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": 1064960,
    "residual_scale": 1.0,
    "vectorizer_mode": "spatial_template",
    "vectorizer_parameters": 5324800
  },
  "args": {
    "a_scale": 1.0,
    "a_warmup_steps": 0,
    "alignment_probe": 32,
    "apical_calibration_mode": "unit_targets",
    "apical_seed": null,
    "augment_train": 1,
    "batch_size": 128,
    "bn_eps": 1e-05,
    "bn_momentum": 0.1,
    "data_dir": "/home/yurenh2/sdrn/data",
    "depth": 56,
    "device": "cuda",
    "epochs": 200,
    "eta_A": 0.01,
    "eta_P": 0.01,
    "eval_every": 0,
    "eval_split": "test",
    "learn_P": 0,
    "loader_seed": 12,
    "lr": 0.03,
    "lr_gamma": 0.1,
    "lr_milestones": "100,150",
    "lr_schedule": "step",
    "max_steps": 0,
    "mirror_batch_size": 1,
    "mirror_eta": 0.1,
    "mirror_every": 16,
    "mirror_noise_std": 1.0,
    "mirror_seed": 3000,
    "mirror_warmup_steps": 0,
    "mode": "dfa",
    "momentum": 0.9,
    "neutral_projection": 0,
    "normalization": "batchnorm",
    "nuisance_scale": 0.0,
    "out": "results/oral_a_dynamic_scaling_v2/dfa_d56_s12.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": 12,
    "split_seed": 2027,
    "traffic_calibration_examples": 64,
    "traffic_ratio": 4.0,
    "traffic_rule": "innovation",
    "traffic_seed": 4000,
    "train_limit": 0,
    "use_residual": 1,
    "val_examples": 0,
    "vectorizer_mode": "spatial_template",
    "warmup_epochs": 0,
    "weight_decay": 0.0001,
    "weight_scale": 1.0,
    "width": 16
  },
  "calibration_metric_space": "full_hidden_field",
  "counters": {
    "apical_warmup_examples": 0,
    "calibration_event_examples": 0,
    "causal_scalar_observations": 0,
    "logical_batch_loss_queries": 0,
    "mirror_conv_examples": 0,
    "mirror_events": 0,
    "mirror_readout_examples": 0,
    "neutral_projection_examples": 0,
    "ordinary_examples": 10000000,
    "per_example_loss_terms": 0,
    "perturbation_events": 0,
    "perturbation_forward_examples": 0,
    "predictor_update_examples": 0,
    "predictor_warmup_examples": 0,
    "traffic_audit_examples": 0,
    "traffic_calibration_examples": 0
  },
  "diagnostics": {
    "early_third_mean": -0.00033107280048878037,
    "innovation_negative_gradient_cosine": [
      -0.001333452295511961,
      0.0001211967901326716,
      0.0003148901159875095,
      -0.0016955535393208265,
      0.0011136706452816725,
      -0.0010295998072251678,
      -8.456464274786413e-05,
      1.7847312847152352e-05,
      -0.00017168925842270255,
      -0.0007532823365181684,
      -0.00022904499201104045,
      -0.0003966555232182145,
      0.0004654741787817329,
      0.0008271206170320511,
      -0.0006747680599801242,
      6.877288979012519e-05,
      -0.0014168135821819305,
      -0.0011028589215129614,
      -0.0001397784799337387,
      0.0001774124102666974,
      0.002049159724265337,
      0.002690684050321579,
      0.0026326305232942104,
      0.0013554745819419622,
      -0.000294134602881968,
      -0.00032216659747064114,
      0.0016200796235352755,
      -0.0013347950298339128,
      0.0017341312486678362,
      0.002476324327290058,
      0.0011687445221468806,
      0.0002642289618961513,
      0.004258967936038971,
      0.0023720012977719307,
      -0.00039734109304845333,
      -0.00044895545579493046,
      0.002411904511973262,
      0.006323819514364004,
      0.008260108530521393,
      0.004710537847131491,
      0.008059056475758553,
      0.004862384404987097,
      0.00503207603469491,
      0.010193176567554474,
      0.004132219590246677,
      0.008396388962864876,
      0.008834676817059517,
      0.0075888317078351974,
      0.006854957435280085,
      0.008016476407647133,
      0.006668945774435997,
      0.00800873339176178,
      0.010383546352386475,
      0.009535139426589012,
      0.01782611943781376
    ],
    "normalization_state": "training_batch_stats_without_running_update",
    "raw_negative_gradient_cosine": [
      -0.001333452295511961,
      0.0001211967901326716,
      0.0003148901159875095,
      -0.0016955535393208265,
      0.0011136706452816725,
      -0.0010295998072251678,
      -8.456464274786413e-05,
      1.7847312847152352e-05,
      -0.00017168925842270255,
      -0.0007532823365181684,
      -0.00022904499201104045,
      -0.0003966555232182145,
      0.0004654741787817329,
      0.0008271206170320511,
      -0.0006747680599801242,
      6.877288979012519e-05,
      -0.0014168135821819305,
      -0.0011028589215129614,
      -0.0001397784799337387,
      0.0001774124102666974,
      0.002049159724265337,
      0.002690684050321579,
      0.0026326305232942104,
      0.0013554745819419622,
      -0.000294134602881968,
      -0.00032216659747064114,
      0.0016200796235352755,
      -0.0013347950298339128,
      0.0017341312486678362,
      0.002476324327290058,
      0.0011687445221468806,
      0.0002642289618961513,
      0.004258967936038971,
      0.0023720012977719307,
      -0.00039734109304845333,
      -0.00044895545579493046,
      0.002411904511973262,
      0.006323819514364004,
      0.008260108530521393,
      0.004710537847131491,
      0.008059056475758553,
      0.004862384404987097,
      0.00503207603469491,
      0.010193176567554474,
      0.004132219590246677,
      0.008396388962864876,
      0.008834676817059517,
      0.0075888317078351974,
      0.006854957435280085,
      0.008016476407647133,
      0.006668945774435997,
      0.00800873339176178,
      0.010383546352386475,
      0.009535139426589012,
      0.01782611943781376
    ],
    "teaching_negative_gradient_cosine": [
      -0.001333452295511961,
      0.0001211967901326716,
      0.0003148901159875095,
      -0.0016955535393208265,
      0.0011136706452816725,
      -0.0010295998072251678,
      -8.456464274786413e-05,
      1.7847312847152352e-05,
      -0.00017168925842270255,
      -0.0007532823365181684,
      -0.00022904499201104045,
      -0.0003966555232182145,
      0.0004654741787817329,
      0.0008271206170320511,
      -0.0006747680599801242,
      6.877288979012519e-05,
      -0.0014168135821819305,
      -0.0011028589215129614,
      -0.0001397784799337387,
      0.0001774124102666974,
      0.002049159724265337,
      0.002690684050321579,
      0.0026326305232942104,
      0.0013554745819419622,
      -0.000294134602881968,
      -0.00032216659747064114,
      0.0016200796235352755,
      -0.0013347950298339128,
      0.0017341312486678362,
      0.002476324327290058,
      0.0011687445221468806,
      0.0002642289618961513,
      0.004258967936038971,
      0.0023720012977719307,
      -0.00039734109304845333,
      -0.00044895545579493046,
      0.002411904511973262,
      0.006323819514364004,
      0.008260108530521393,
      0.004710537847131491,
      0.008059056475758553,
      0.004862384404987097,
      0.00503207603469491,
      0.010193176567554474,
      0.004132219590246677,
      0.008396388962864876,
      0.008834676817059517,
      0.0075888317078351974,
      0.006854957435280085,
      0.008016476407647133,
      0.006668945774435997,
      0.00800873339176178,
      0.010383546352386475,
      0.009535139426589012,
      0.01782611943781376
    ],
    "wall_s": 0.20418858528137207
  },
  "epochs": [
    {
      "epoch": 1,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 391,
      "train_examples": 50000,
      "train_loss": 20.94000764160156
    },
    {
      "epoch": 2,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 782,
      "train_examples": 50000,
      "train_loss": 19.62649717285156
    },
    {
      "epoch": 3,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 1173,
      "train_examples": 50000,
      "train_loss": 25.090193415527345
    },
    {
      "epoch": 4,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 1564,
      "train_examples": 50000,
      "train_loss": 29.649140238647462
    },
    {
      "epoch": 5,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 1955,
      "train_examples": 50000,
      "train_loss": 34.545018566894534
    },
    {
      "epoch": 6,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 2346,
      "train_examples": 50000,
      "train_loss": 48.3627574975586
    },
    {
      "epoch": 7,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 2737,
      "train_examples": 50000,
      "train_loss": 77.06149560791016
    },
    {
      "epoch": 8,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 3128,
      "train_examples": 50000,
      "train_loss": 101.49769426757813
    },
    {
      "epoch": 9,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 3519,
      "train_examples": 50000,
      "train_loss": 150.34198771972657
    },
    {
      "epoch": 10,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 3910,
      "train_examples": 50000,
      "train_loss": 177.8562303808594
    },
    {
      "epoch": 11,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 4301,
      "train_examples": 50000,
      "train_loss": 221.7907431298828
    },
    {
      "epoch": 12,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 4692,
      "train_examples": 50000,
      "train_loss": 294.5412326269531
    },
    {
      "epoch": 13,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 5083,
      "train_examples": 50000,
      "train_loss": 344.3779750390625
    },
    {
      "epoch": 14,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 5474,
      "train_examples": 50000,
      "train_loss": 411.1375525390625
    },
    {
      "epoch": 15,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 5865,
      "train_examples": 50000,
      "train_loss": 560.7590898242188
    },
    {
      "epoch": 16,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 6256,
      "train_examples": 50000,
      "train_loss": 623.2798921875
    },
    {
      "epoch": 17,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 6647,
      "train_examples": 50000,
      "train_loss": 752.5977274804687
    },
    {
      "epoch": 18,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 7038,
      "train_examples": 50000,
      "train_loss": 1002.7074587304687
    },
    {
      "epoch": 19,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 7429,
      "train_examples": 50000,
      "train_loss": 1313.94790640625
    },
    {
      "epoch": 20,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 7820,
      "train_examples": 50000,
      "train_loss": 1414.9425067773439
    },
    {
      "epoch": 21,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 8211,
      "train_examples": 50000,
      "train_loss": 1503.092805625
    },
    {
      "epoch": 22,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 8602,
      "train_examples": 50000,
      "train_loss": 1937.405467265625
    },
    {
      "epoch": 23,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 8993,
      "train_examples": 50000,
      "train_loss": 2087.423003203125
    },
    {
      "epoch": 24,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 9384,
      "train_examples": 50000,
      "train_loss": 2564.388800546875
    },
    {
      "epoch": 25,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 9775,
      "train_examples": 50000,
      "train_loss": 2609.4218796875
    },
    {
      "epoch": 26,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 10166,
      "train_examples": 50000,
      "train_loss": 3110.83165625
    },
    {
      "epoch": 27,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 10557,
      "train_examples": 50000,
      "train_loss": 3776.0829210546876
    },
    {
      "epoch": 28,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 10948,
      "train_examples": 50000,
      "train_loss": 3880.975733046875
    },
    {
      "epoch": 29,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 11339,
      "train_examples": 50000,
      "train_loss": 4387.7525015625
    },
    {
      "epoch": 30,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 11730,
      "train_examples": 50000,
      "train_loss": 4984.768659375
    },
    {
      "epoch": 31,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 12121,
      "train_examples": 50000,
      "train_loss": 5330.03115265625
    },
    {
      "epoch": 32,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 12512,
      "train_examples": 50000,
      "train_loss": 5846.50104453125
    },
    {
      "epoch": 33,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 12903,
      "train_examples": 50000,
      "train_loss": 5710.3653346875
    },
    {
      "epoch": 34,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 13294,
      "train_examples": 50000,
      "train_loss": 6278.23847453125
    },
    {
      "epoch": 35,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 13685,
      "train_examples": 50000,
      "train_loss": 6830.864655
    },
    {
      "epoch": 36,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 14076,
      "train_examples": 50000,
      "train_loss": 6546.9267015625
    },
    {
      "epoch": 37,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 14467,
      "train_examples": 50000,
      "train_loss": 7391.49460296875
    },
    {
      "epoch": 38,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 14858,
      "train_examples": 50000,
      "train_loss": 8033.417875
    },
    {
      "epoch": 39,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 15249,
      "train_examples": 50000,
      "train_loss": 8922.7807009375
    },
    {
      "epoch": 40,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 15640,
      "train_examples": 50000,
      "train_loss": 8632.656355625
    },
    {
      "epoch": 41,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 16031,
      "train_examples": 50000,
      "train_loss": 9130.92773734375
    },
    {
      "epoch": 42,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 16422,
      "train_examples": 50000,
      "train_loss": 10076.9714134375
    },
    {
      "epoch": 43,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 16813,
      "train_examples": 50000,
      "train_loss": 11774.00065875
    },
    {
      "epoch": 44,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 17204,
      "train_examples": 50000,
      "train_loss": 11593.69164125
    },
    {
      "epoch": 45,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 17595,
      "train_examples": 50000,
      "train_loss": 11887.653268125
    },
    {
      "epoch": 46,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 17986,
      "train_examples": 50000,
      "train_loss": 12574.010313125
    },
    {
      "epoch": 47,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 18377,
      "train_examples": 50000,
      "train_loss": 14421.15337375
    },
    {
      "epoch": 48,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 18768,
      "train_examples": 50000,
      "train_loss": 13520.107968125
    },
    {
      "epoch": 49,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 19159,
      "train_examples": 50000,
      "train_loss": 15746.6091353125
    },
    {
      "epoch": 50,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 19550,
      "train_examples": 50000,
      "train_loss": 15733.638566875
    },
    {
      "epoch": 51,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 19941,
      "train_examples": 50000,
      "train_loss": 18329.241820625
    },
    {
      "epoch": 52,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 20332,
      "train_examples": 50000,
      "train_loss": 18431.455229375
    },
    {
      "epoch": 53,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 20723,
      "train_examples": 50000,
      "train_loss": 17570.81183625
    },
    {
      "epoch": 54,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 21114,
      "train_examples": 50000,
      "train_loss": 16973.86219875
    },
    {
      "epoch": 55,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 21505,
      "train_examples": 50000,
      "train_loss": 19579.385968125
    },
    {
      "epoch": 56,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 21896,
      "train_examples": 50000,
      "train_loss": 23204.30533
    },
    {
      "epoch": 57,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 22287,
      "train_examples": 50000,
      "train_loss": 20706.27900125
    },
    {
      "epoch": 58,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 22678,
      "train_examples": 50000,
      "train_loss": 21582.72516875
    },
    {
      "epoch": 59,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 23069,
      "train_examples": 50000,
      "train_loss": 26746.427590625
    },
    {
      "epoch": 60,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 23460,
      "train_examples": 50000,
      "train_loss": 26144.13405625
    },
    {
      "epoch": 61,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 23851,
      "train_examples": 50000,
      "train_loss": 24698.052254375
    },
    {
      "epoch": 62,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 24242,
      "train_examples": 50000,
      "train_loss": 23851.24748625
    },
    {
      "epoch": 63,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 24633,
      "train_examples": 50000,
      "train_loss": 27929.38762125
    },
    {
      "epoch": 64,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 25024,
      "train_examples": 50000,
      "train_loss": 30175.7901
    },
    {
      "epoch": 65,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 25415,
      "train_examples": 50000,
      "train_loss": 28197.3008525
    },
    {
      "epoch": 66,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 25806,
      "train_examples": 50000,
      "train_loss": 29571.4402075
    },
    {
      "epoch": 67,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 26197,
      "train_examples": 50000,
      "train_loss": 31887.56947375
    },
    {
      "epoch": 68,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 26588,
      "train_examples": 50000,
      "train_loss": 31777.02970375
    },
    {
      "epoch": 69,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 26979,
      "train_examples": 50000,
      "train_loss": 33884.86967375
    },
    {
      "epoch": 70,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 27370,
      "train_examples": 50000,
      "train_loss": 39072.53294125
    },
    {
      "epoch": 71,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 27761,
      "train_examples": 50000,
      "train_loss": 37428.5231675
    },
    {
      "epoch": 72,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 28152,
      "train_examples": 50000,
      "train_loss": 36754.15567
    },
    {
      "epoch": 73,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 28543,
      "train_examples": 50000,
      "train_loss": 40484.509706875
    },
    {
      "epoch": 74,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 28934,
      "train_examples": 50000,
      "train_loss": 41803.5729125
    },
    {
      "epoch": 75,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 29325,
      "train_examples": 50000,
      "train_loss": 35832.9667425
    },
    {
      "epoch": 76,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 29716,
      "train_examples": 50000,
      "train_loss": 40203.92737625
    },
    {
      "epoch": 77,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 30107,
      "train_examples": 50000,
      "train_loss": 40912.255264375
    },
    {
      "epoch": 78,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 30498,
      "train_examples": 50000,
      "train_loss": 43315.66097625
    },
    {
      "epoch": 79,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 30889,
      "train_examples": 50000,
      "train_loss": 43469.07302875
    },
    {
      "epoch": 80,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 31280,
      "train_examples": 50000,
      "train_loss": 43828.5124
    },
    {
      "epoch": 81,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 31671,
      "train_examples": 50000,
      "train_loss": 45599.000315
    },
    {
      "epoch": 82,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 32062,
      "train_examples": 50000,
      "train_loss": 40848.565963125
    },
    {
      "epoch": 83,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 32453,
      "train_examples": 50000,
      "train_loss": 50698.45087125
    },
    {
      "epoch": 84,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 32844,
      "train_examples": 50000,
      "train_loss": 45765.55349375
    },
    {
      "epoch": 85,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 33235,
      "train_examples": 50000,
      "train_loss": 47759.3486175
    },
    {
      "epoch": 86,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 33626,
      "train_examples": 50000,
      "train_loss": 47235.41903
    },
    {
      "epoch": 87,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 34017,
      "train_examples": 50000,
      "train_loss": 55139.881675
    },
    {
      "epoch": 88,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 34408,
      "train_examples": 50000,
      "train_loss": 54489.0964925
    },
    {
      "epoch": 89,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 34799,
      "train_examples": 50000,
      "train_loss": 50621.15795
    },
    {
      "epoch": 90,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 35190,
      "train_examples": 50000,
      "train_loss": 52931.92026625
    },
    {
      "epoch": 91,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 35581,
      "train_examples": 50000,
      "train_loss": 61464.17159625
    },
    {
      "epoch": 92,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 35972,
      "train_examples": 50000,
      "train_loss": 58305.407645
    },
    {
      "epoch": 93,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 36363,
      "train_examples": 50000,
      "train_loss": 57352.33292125
    },
    {
      "epoch": 94,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 36754,
      "train_examples": 50000,
      "train_loss": 58644.4164275
    },
    {
      "epoch": 95,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 37145,
      "train_examples": 50000,
      "train_loss": 62408.0420275
    },
    {
      "epoch": 96,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 37536,
      "train_examples": 50000,
      "train_loss": 61532.5740675
    },
    {
      "epoch": 97,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 37927,
      "train_examples": 50000,
      "train_loss": 59007.93284375
    },
    {
      "epoch": 98,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 38318,
      "train_examples": 50000,
      "train_loss": 68775.532
    },
    {
      "epoch": 99,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 38709,
      "train_examples": 50000,
      "train_loss": 64542.84597625
    },
    {
      "epoch": 100,
      "lr": 0.03,
      "output_lr": 0.1,
      "step": 39100,
      "train_examples": 50000,
      "train_loss": 70629.09657
    },
    {
      "epoch": 101,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 39491,
      "train_examples": 50000,
      "train_loss": 14654.2372521875
    },
    {
      "epoch": 102,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 39882,
      "train_examples": 50000,
      "train_loss": 10468.6515740625
    },
    {
      "epoch": 103,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 40273,
      "train_examples": 50000,
      "train_loss": 9067.7611084375
    },
    {
      "epoch": 104,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 40664,
      "train_examples": 50000,
      "train_loss": 8423.380498125
    },
    {
      "epoch": 105,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 41055,
      "train_examples": 50000,
      "train_loss": 8158.55437375
    },
    {
      "epoch": 106,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 41446,
      "train_examples": 50000,
      "train_loss": 7619.72909359375
    },
    {
      "epoch": 107,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 41837,
      "train_examples": 50000,
      "train_loss": 7046.0333971875
    },
    {
      "epoch": 108,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 42228,
      "train_examples": 50000,
      "train_loss": 6775.8592484375
    },
    {
      "epoch": 109,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 42619,
      "train_examples": 50000,
      "train_loss": 6478.0563859375
    },
    {
      "epoch": 110,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 43010,
      "train_examples": 50000,
      "train_loss": 6159.16063015625
    },
    {
      "epoch": 111,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 43401,
      "train_examples": 50000,
      "train_loss": 5799.35488171875
    },
    {
      "epoch": 112,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 43792,
      "train_examples": 50000,
      "train_loss": 5433.1685890625
    },
    {
      "epoch": 113,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 44183,
      "train_examples": 50000,
      "train_loss": 5350.97988546875
    },
    {
      "epoch": 114,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 44574,
      "train_examples": 50000,
      "train_loss": 4986.5760428125
    },
    {
      "epoch": 115,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 44965,
      "train_examples": 50000,
      "train_loss": 5325.34049234375
    },
    {
      "epoch": 116,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 45356,
      "train_examples": 50000,
      "train_loss": 5104.84509546875
    },
    {
      "epoch": 117,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 45747,
      "train_examples": 50000,
      "train_loss": 5064.1554721875
    },
    {
      "epoch": 118,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 46138,
      "train_examples": 50000,
      "train_loss": 5048.13006140625
    },
    {
      "epoch": 119,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 46529,
      "train_examples": 50000,
      "train_loss": 4887.15388859375
    },
    {
      "epoch": 120,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 46920,
      "train_examples": 50000,
      "train_loss": 4995.7025578125
    },
    {
      "epoch": 121,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 47311,
      "train_examples": 50000,
      "train_loss": 5238.9835584375
    },
    {
      "epoch": 122,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 47702,
      "train_examples": 50000,
      "train_loss": 4834.56711734375
    },
    {
      "epoch": 123,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 48093,
      "train_examples": 50000,
      "train_loss": 5015.1136084375
    },
    {
      "epoch": 124,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 48484,
      "train_examples": 50000,
      "train_loss": 5329.49159671875
    },
    {
      "epoch": 125,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 48875,
      "train_examples": 50000,
      "train_loss": 5179.48424609375
    },
    {
      "epoch": 126,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 49266,
      "train_examples": 50000,
      "train_loss": 4766.6652128125
    },
    {
      "epoch": 127,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 49657,
      "train_examples": 50000,
      "train_loss": 5021.14735359375
    },
    {
      "epoch": 128,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 50048,
      "train_examples": 50000,
      "train_loss": 5365.06995828125
    },
    {
      "epoch": 129,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 50439,
      "train_examples": 50000,
      "train_loss": 4384.51806296875
    },
    {
      "epoch": 130,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 50830,
      "train_examples": 50000,
      "train_loss": 5167.5980784375
    },
    {
      "epoch": 131,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 51221,
      "train_examples": 50000,
      "train_loss": 5362.407710546875
    },
    {
      "epoch": 132,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 51612,
      "train_examples": 50000,
      "train_loss": 4922.89476328125
    },
    {
      "epoch": 133,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 52003,
      "train_examples": 50000,
      "train_loss": 5145.946841796875
    },
    {
      "epoch": 134,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 52394,
      "train_examples": 50000,
      "train_loss": 4707.67387203125
    },
    {
      "epoch": 135,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 52785,
      "train_examples": 50000,
      "train_loss": 5891.3174746875
    },
    {
      "epoch": 136,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 53176,
      "train_examples": 50000,
      "train_loss": 4795.557129375
    },
    {
      "epoch": 137,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 53567,
      "train_examples": 50000,
      "train_loss": 4844.1970709375
    },
    {
      "epoch": 138,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 53958,
      "train_examples": 50000,
      "train_loss": 5076.74701140625
    },
    {
      "epoch": 139,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 54349,
      "train_examples": 50000,
      "train_loss": 5042.289148828125
    },
    {
      "epoch": 140,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 54740,
      "train_examples": 50000,
      "train_loss": 5365.8960978125
    },
    {
      "epoch": 141,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 55131,
      "train_examples": 50000,
      "train_loss": 5071.8902071875
    },
    {
      "epoch": 142,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 55522,
      "train_examples": 50000,
      "train_loss": 5137.35602296875
    },
    {
      "epoch": 143,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 55913,
      "train_examples": 50000,
      "train_loss": 5353.469021875
    },
    {
      "epoch": 144,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 56304,
      "train_examples": 50000,
      "train_loss": 5375.27371984375
    },
    {
      "epoch": 145,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 56695,
      "train_examples": 50000,
      "train_loss": 5149.46602109375
    },
    {
      "epoch": 146,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 57086,
      "train_examples": 50000,
      "train_loss": 5708.603248125
    },
    {
      "epoch": 147,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 57477,
      "train_examples": 50000,
      "train_loss": 5485.7706896875
    },
    {
      "epoch": 148,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 57868,
      "train_examples": 50000,
      "train_loss": 5218.4024771875
    },
    {
      "epoch": 149,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 58259,
      "train_examples": 50000,
      "train_loss": 5457.35446234375
    },
    {
      "epoch": 150,
      "lr": 0.003,
      "output_lr": 0.010000000000000002,
      "step": 58650,
      "train_examples": 50000,
      "train_loss": 5654.4977634375
    },
    {
      "epoch": 151,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 59041,
      "train_examples": 50000,
      "train_loss": 1917.4399200390626
    },
    {
      "epoch": 152,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 59432,
      "train_examples": 50000,
      "train_loss": 1666.58332859375
    },
    {
      "epoch": 153,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 59823,
      "train_examples": 50000,
      "train_loss": 1537.5342199609374
    },
    {
      "epoch": 154,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 60214,
      "train_examples": 50000,
      "train_loss": 1442.5484925390624
    },
    {
      "epoch": 155,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 60605,
      "train_examples": 50000,
      "train_loss": 1339.382270703125
    },
    {
      "epoch": 156,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 60996,
      "train_examples": 50000,
      "train_loss": 1298.7273259375
    },
    {
      "epoch": 157,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 61387,
      "train_examples": 50000,
      "train_loss": 1237.4698423242187
    },
    {
      "epoch": 158,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 61778,
      "train_examples": 50000,
      "train_loss": 1184.8028786328125
    },
    {
      "epoch": 159,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 62169,
      "train_examples": 50000,
      "train_loss": 1164.850845234375
    },
    {
      "epoch": 160,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 62560,
      "train_examples": 50000,
      "train_loss": 1108.2915394140625
    },
    {
      "epoch": 161,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 62951,
      "train_examples": 50000,
      "train_loss": 1075.1936887890624
    },
    {
      "epoch": 162,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 63342,
      "train_examples": 50000,
      "train_loss": 1055.1205018554688
    },
    {
      "epoch": 163,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 63733,
      "train_examples": 50000,
      "train_loss": 1035.362221640625
    },
    {
      "epoch": 164,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 64124,
      "train_examples": 50000,
      "train_loss": 988.9151735546875
    },
    {
      "epoch": 165,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 64515,
      "train_examples": 50000,
      "train_loss": 993.6608714257812
    },
    {
      "epoch": 166,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 64906,
      "train_examples": 50000,
      "train_loss": 960.6094983984375
    },
    {
      "epoch": 167,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 65297,
      "train_examples": 50000,
      "train_loss": 925.7214187695313
    },
    {
      "epoch": 168,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 65688,
      "train_examples": 50000,
      "train_loss": 945.243518203125
    },
    {
      "epoch": 169,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 66079,
      "train_examples": 50000,
      "train_loss": 915.3768917578125
    },
    {
      "epoch": 170,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 66470,
      "train_examples": 50000,
      "train_loss": 900.5765274609375
    },
    {
      "epoch": 171,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 66861,
      "train_examples": 50000,
      "train_loss": 891.4864741992187
    },
    {
      "epoch": 172,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 67252,
      "train_examples": 50000,
      "train_loss": 869.0761085742188
    },
    {
      "epoch": 173,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 67643,
      "train_examples": 50000,
      "train_loss": 884.9851100195312
    },
    {
      "epoch": 174,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 68034,
      "train_examples": 50000,
      "train_loss": 876.4683806445313
    },
    {
      "epoch": 175,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 68425,
      "train_examples": 50000,
      "train_loss": 848.3266902734375
    },
    {
      "epoch": 176,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 68816,
      "train_examples": 50000,
      "train_loss": 845.732801015625
    },
    {
      "epoch": 177,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 69207,
      "train_examples": 50000,
      "train_loss": 839.8744810351562
    },
    {
      "epoch": 178,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 69598,
      "train_examples": 50000,
      "train_loss": 845.5655024609375
    },
    {
      "epoch": 179,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 69989,
      "train_examples": 50000,
      "train_loss": 824.5605144726562
    },
    {
      "epoch": 180,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 70380,
      "train_examples": 50000,
      "train_loss": 834.7904227929688
    },
    {
      "epoch": 181,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 70771,
      "train_examples": 50000,
      "train_loss": 850.2991547460938
    },
    {
      "epoch": 182,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 71162,
      "train_examples": 50000,
      "train_loss": 820.90386671875
    },
    {
      "epoch": 183,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 71553,
      "train_examples": 50000,
      "train_loss": 827.5199787695312
    },
    {
      "epoch": 184,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 71944,
      "train_examples": 50000,
      "train_loss": 832.7114706640625
    },
    {
      "epoch": 185,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 72335,
      "train_examples": 50000,
      "train_loss": 800.0516747265625
    },
    {
      "epoch": 186,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 72726,
      "train_examples": 50000,
      "train_loss": 794.9772009375
    },
    {
      "epoch": 187,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 73117,
      "train_examples": 50000,
      "train_loss": 802.980529453125
    },
    {
      "epoch": 188,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 73508,
      "train_examples": 50000,
      "train_loss": 817.6020956640625
    },
    {
      "epoch": 189,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 73899,
      "train_examples": 50000,
      "train_loss": 798.885476328125
    },
    {
      "epoch": 190,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 74290,
      "train_examples": 50000,
      "train_loss": 825.81932140625
    },
    {
      "epoch": 191,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 74681,
      "train_examples": 50000,
      "train_loss": 789.4225649414062
    },
    {
      "epoch": 192,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 75072,
      "train_examples": 50000,
      "train_loss": 817.1468815625
    },
    {
      "epoch": 193,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 75463,
      "train_examples": 50000,
      "train_loss": 803.5091401757812
    },
    {
      "epoch": 194,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 75854,
      "train_examples": 50000,
      "train_loss": 799.3935744921876
    },
    {
      "epoch": 195,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 76245,
      "train_examples": 50000,
      "train_loss": 781.3792027734376
    },
    {
      "epoch": 196,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 76636,
      "train_examples": 50000,
      "train_loss": 796.2545483984375
    },
    {
      "epoch": 197,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 77027,
      "train_examples": 50000,
      "train_loss": 851.6191300195312
    },
    {
      "epoch": 198,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 77418,
      "train_examples": 50000,
      "train_loss": 756.63056
    },
    {
      "epoch": 199,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 77809,
      "train_examples": 50000,
      "train_loss": 822.309709140625
    },
    {
      "epoch": 200,
      "lr": 0.00030000000000000003,
      "output_lr": 0.0010000000000000002,
      "step": 78200,
      "train_examples": 50000,
      "train_loss": 765.2383037109375
    }
  ],
  "evaluation_protocol": {
    "test_evaluations": 1,
    "test_used_for_selection": false,
    "validation_evaluations": 0
  },
  "final": {
    "accuracy": 0.3382,
    "epoch": 200,
    "evaluation_split": "test",
    "finite": true,
    "loss": 606.93005625,
    "step": 78200
  },
  "hardware": {
    "cuda_device_name": "NVIDIA GeForce GTX 1080",
    "cuda_visible_devices": "7",
    "device": "cuda",
    "device_total_memory_bytes": 8507949056,
    "peak_memory_allocated_bytes": 3339442688,
    "peak_memory_reserved_bytes": 4676648960,
    "torch_version": "2.3.1+cu118"
  },
  "protocol_family": "oral_a_cifar_local_resnet_development",
  "provenance": {
    "git_commit": "8f9364e8ee2e2a19eb83b58452fa0e8cb155353c",
    "git_tracked_dirty": false
  },
  "schema_version": 1,
  "split": {
    "cifar_source_files": [
      {
        "bytes": 31035704,
        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_1",
        "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
      },
      {
        "bytes": 31035320,
        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_2",
        "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
      },
      {
        "bytes": 31035999,
        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_3",
        "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
      },
      {
        "bytes": 31035696,
        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_4",
        "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
      },
      {
        "bytes": 31035623,
        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_5",
        "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
      },
      {
        "bytes": 31035526,
        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/test_batch",
        "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831"
      }
    ],
    "dataset": "cifar10",
    "evaluation_split": "test",
    "input_layout": "NCHW",
    "input_shape": [
      3,
      32,
      32
    ],
    "loader_seed": 12,
    "normalization_mean": [
      0.49140000343322754,
      0.4821999967098236,
      0.4465000033378601
    ],
    "normalization_std": [
      0.24699999392032623,
      0.2434999942779541,
      0.26159998774528503
    ],
    "split_from_training_only": true,
    "split_seed": null,
    "test_examples": 10000,
    "train_examples": 50000,
    "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
    "validation_class_counts": {},
    "validation_examples": 0,
    "validation_index_sha256": null
  },
  "timing": {
    "apical_warmup_wall_s": 0.0,
    "evaluation_wall_s": 2.0357413291931152,
    "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": 10408.400108575821,
    "train_wall_s": 10406.33801150322
  },
  "work": {
    "apical_macs_per_example": 5324800,
    "causal_scalar_observations": 0,
    "components": {
      "apical_projection_macs": 53248000000000,
      "apical_regression_macs": 0,
      "bp_reverse_macs_estimate": 0,
      "kp_reciprocal_correlation_macs": 0,
      "local_weight_correlation_macs": 1254856960000000,
      "mirror_feedback_prediction_macs": 0,
      "mirror_local_correlation_macs": 0,
      "mirror_response_macs": 0,
      "ordinary_forward_macs": 1254856960000000,
      "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": 10000000,
    "total_forward_equivalent_examples": 10000000,
    "total_macs_estimate": 2562961920000000
  }
}