{
  "architecture": {
    "adaptive_apical_parameters": 0,
    "adaptive_feedback_parameters": 848512,
    "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": 0,
    "forward_parameters": 853018,
    "hidden_shapes": [
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        16,
        32,
        32
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        32,
        16,
        16
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ],
      [
        64,
        8,
        8
      ]
    ],
    "normalization": "batchnorm",
    "predictor_parameters": 0,
    "residual_scale": 1.0,
    "vectorizer_mode": null,
    "vectorizer_parameters": 0
  },
  "args": {
    "a_scale": 1.0,
    "a_warmup_steps": 0,
    "alignment_probe": 32,
    "apical_calibration_mode": "unit_targets",
    "apical_seed": null,
    "augment_train": 1,
    "batch_size": 128,
    "bn_eps": 1e-05,
    "bn_momentum": 0.1,
    "data_dir": "/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": 10,
    "lr": 0.1,
    "lr_gamma": 0.1,
    "lr_milestones": "100,150",
    "lr_schedule": "step",
    "max_steps": 0,
    "mirror_batch_size": 1,
    "mirror_eta": 0.1,
    "mirror_every": 16,
    "mirror_noise_std": 1.0,
    "mirror_seed": 3000,
    "mirror_warmup_steps": 0,
    "mode": "kp",
    "momentum": 0.9,
    "neutral_projection": 0,
    "normalization": "batchnorm",
    "nuisance_scale": 0.0,
    "out": "results/oral_a_dynamic_scaling_v2/clean_kp_d56_s10.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": 10,
    "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": "reciprocal_local_activity_products",
  "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": 0.9994567665788863,
    "feedback_forward_cosine": [
      0.9995390176773071,
      0.9994896054267883,
      0.9993683695793152,
      0.9992607831954956,
      0.9991724491119385,
      0.9991496801376343,
      0.9994419813156128,
      0.9994545578956604,
      0.9993976354598999,
      0.9993428587913513,
      0.9996061325073242,
      0.9995467662811279,
      0.9996442794799805,
      0.9996175765991211,
      0.9997511506080627,
      0.9997085332870483,
      0.9997986555099487,
      0.9996837973594666,
      0.9996693134307861,
      0.9997653961181641,
      0.9995806813240051,
      0.99955153465271,
      0.9996522665023804,
      0.999573826789856,
      0.9996734857559204,
      0.9995680451393127,
      0.9996894598007202,
      0.9995835423469543,
      0.9996563196182251,
      0.9995564222335815,
      0.9996431469917297,
      0.9995015859603882,
      0.9996466636657715,
      0.9994857311248779,
      0.9995725154876709,
      0.9993917942047119,
      0.9997338056564331,
      0.9998142719268799,
      0.9997738003730774,
      0.9997599124908447,
      0.9997965097427368,
      0.9997284412384033,
      0.9997711181640625,
      0.9996548891067505,
      0.999747097492218,
      0.9995355010032654,
      0.9995253086090088,
      0.9992527961730957,
      0.9991543292999268,
      0.9988086819648743,
      0.9985666871070862,
      0.9980965256690979,
      0.9990485310554504,
      0.9984483122825623,
      0.9999804496765137
    ],
    "feedback_forward_norm_ratio": [
      0.9986742734909058,
      1.0006389617919922,
      0.9969250559806824,
      1.0015281438827515,
      0.9994803667068481,
      1.0023949146270752,
      0.9992635250091553,
      1.0001678466796875,
      0.9981926679611206,
      1.0019519329071045,
      0.9987691044807434,
      1.0017606019973755,
      0.9995517730712891,
      1.0003198385238647,
      0.9993669390678406,
      1.0014435052871704,
      0.9996762871742249,
      1.0010688304901123,
      0.9983429312705994,
      1.000935435295105,
      0.9987174272537231,
      1.001394271850586,
      0.9991419315338135,
      1.0014270544052124,
      0.999220609664917,
      1.001044511795044,
      0.9996599555015564,
      1.0019431114196777,
      0.999127209186554,
      1.0018210411071777,
      0.9987342357635498,
      1.002028465270996,
      0.999176025390625,
      1.0014485120773315,
      0.9987326860427856,
      1.0022226572036743,
      0.999302089214325,
      1.0003089904785156,
      0.9998563528060913,
      1.0008293390274048,
      0.9998266696929932,
      1.000866413116455,
      0.9995371103286743,
      1.0013400316238403,
      0.9996753334999084,
      1.0014046430587769,
      0.9987162947654724,
      1.0021916627883911,
      0.9979419708251953,
      1.002637267112732,
      0.9971224069595337,
      1.0037763118743896,
      0.9985743761062622,
      1.0024051666259766,
      1.0020339488983154
    ],
    "feedback_forward_relative_error": [
      0.030373815447092056,
      0.031964294612407684,
      0.03562188893556595,
      0.03850705549120903,
      0.04067551717162132,
      0.04135864973068237,
      0.03340218961238861,
      0.03303094580769539,
      0.03472533077001572,
      0.03634229302406311,
      0.028073109686374664,
      0.030188247561454773,
      0.026667024940252304,
      0.02766183763742447,
      0.02230963297188282,
      0.02419978752732277,
      0.020063288509845734,
      0.025182627141475677,
      0.025750935077667236,
      0.021695468574762344,
      0.02896524965763092,
      0.029999585822224617,
      0.02637401595711708,
      0.029252730309963226,
      0.025553440675139427,
      0.02942599356174469,
      0.024919671937823296,
      0.02895422838628292,
      0.026219619438052177,
      0.029866602271795273,
      0.02672802284359932,
      0.03166596591472626,
      0.02658591978251934,
      0.03212727606296539,
      0.029248615726828575,
      0.03498656675219536,
      0.023075399920344353,
      0.019277160987257957,
      0.0212686900049448,
      0.021937349811196327,
      0.020174607634544373,
      0.023328101262450218,
      0.021395066753029823,
      0.026319433003664017,
      0.02248847484588623,
      0.030529577285051346,
      0.03081503137946129,
      0.038761161267757416,
      0.04113198071718216,
      0.0489472858607769,
      0.053541116416454315,
      0.06193094700574875,
      0.04361523687839508,
      0.05582822114229202,
      0.006581505760550499
    ],
    "innovation_negative_gradient_cosine": [
      0.999534547328949,
      0.999441385269165,
      0.9995296001434326,
      0.9990184307098389,
      0.9995428323745728,
      0.9992603063583374,
      0.9995453953742981,
      0.9992389678955078,
      0.9995508193969727,
      0.9993209838867188,
      0.99955815076828,
      0.9994360208511353,
      0.9995626211166382,
      0.999459981918335,
      0.9995666742324829,
      0.9995467662811279,
      0.9995676875114441,
      0.9995406270027161,
      0.9995849132537842,
      0.9995675086975098,
      0.9996196031570435,
      0.999435544013977,
      0.9996318817138672,
      0.9994937181472778,
      0.9996384382247925,
      0.9995115995407104,
      0.9996570348739624,
      0.9994932413101196,
      0.9996674060821533,
      0.9994916915893555,
      0.9996871948242188,
      0.9995210766792297,
      0.9997069239616394,
      0.9995291829109192,
      0.9997122883796692,
      0.9995144605636597,
      0.9997189044952393,
      0.9997336864471436,
      0.9997418522834778,
      0.9996910095214844,
      0.9997549057006836,
      0.999707043170929,
      0.9997638463973999,
      0.9997535347938538,
      0.9997890591621399,
      0.9998000860214233,
      0.9997377395629883,
      0.9997023344039917,
      0.9997260570526123,
      0.999669075012207,
      0.9996702671051025,
      0.9995840787887573,
      0.9998461008071899,
      0.9998620748519897,
      0.6710851192474365
    ],
    "mean_feedback_forward_cosine": 0.9994896823709661,
    "mean_feedback_forward_relative_error": 0.03053843260488727,
    "normalization_state": "training_batch_stats_without_running_update",
    "raw_negative_gradient_cosine": [
      0.999534547328949,
      0.999441385269165,
      0.9995296001434326,
      0.9990184307098389,
      0.9995428323745728,
      0.9992603063583374,
      0.9995453953742981,
      0.9992389678955078,
      0.9995508193969727,
      0.9993209838867188,
      0.99955815076828,
      0.9994360208511353,
      0.9995626211166382,
      0.999459981918335,
      0.9995666742324829,
      0.9995467662811279,
      0.9995676875114441,
      0.9995406270027161,
      0.9995849132537842,
      0.9995675086975098,
      0.9996196031570435,
      0.999435544013977,
      0.9996318817138672,
      0.9994937181472778,
      0.9996384382247925,
      0.9995115995407104,
      0.9996570348739624,
      0.9994932413101196,
      0.9996674060821533,
      0.9994916915893555,
      0.9996871948242188,
      0.9995210766792297,
      0.9997069239616394,
      0.9995291829109192,
      0.9997122883796692,
      0.9995144605636597,
      0.9997189044952393,
      0.9997336864471436,
      0.9997418522834778,
      0.9996910095214844,
      0.9997549057006836,
      0.999707043170929,
      0.9997638463973999,
      0.9997535347938538,
      0.9997890591621399,
      0.9998000860214233,
      0.9997377395629883,
      0.9997023344039917,
      0.9997260570526123,
      0.999669075012207,
      0.9996702671051025,
      0.9995840787887573,
      0.9998461008071899,
      0.9998620748519897,
      0.6710851192474365
    ],
    "teaching_negative_gradient_cosine": [
      0.999534547328949,
      0.999441385269165,
      0.9995296001434326,
      0.9990184307098389,
      0.9995428323745728,
      0.9992603063583374,
      0.9995453953742981,
      0.9992389678955078,
      0.9995508193969727,
      0.9993209838867188,
      0.99955815076828,
      0.9994360208511353,
      0.9995626211166382,
      0.999459981918335,
      0.9995666742324829,
      0.9995467662811279,
      0.9995676875114441,
      0.9995406270027161,
      0.9995849132537842,
      0.9995675086975098,
      0.9996196031570435,
      0.999435544013977,
      0.9996318817138672,
      0.9994937181472778,
      0.9996384382247925,
      0.9995115995407104,
      0.9996570348739624,
      0.9994932413101196,
      0.9996674060821533,
      0.9994916915893555,
      0.9996871948242188,
      0.9995210766792297,
      0.9997069239616394,
      0.9995291829109192,
      0.9997122883796692,
      0.9995144605636597,
      0.9997189044952393,
      0.9997336864471436,
      0.9997418522834778,
      0.9996910095214844,
      0.9997549057006836,
      0.999707043170929,
      0.9997638463973999,
      0.9997535347938538,
      0.9997890591621399,
      0.9998000860214233,
      0.9997377395629883,
      0.9997023344039917,
      0.9997260570526123,
      0.999669075012207,
      0.9996702671051025,
      0.9995840787887573,
      0.9998461008071899,
      0.9998620748519897,
      0.6710851192474365
    ],
    "wall_s": 0.1840651035308838
  },
  "epochs": [
    {
      "epoch": 1,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3464447259902954,
        "mean_feedback_forward_cosine": 0.29027721380645577,
        "mean_feedback_forward_relative_error": 1.178256191448732,
        "min_feedback_forward_cosine": 0.09826618432998657
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 391,
      "train_examples": 50000,
      "train_loss": 2.3704615573501586
    },
    {
      "epoch": 2,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.342604398727417,
        "mean_feedback_forward_cosine": 0.31901492232626133,
        "mean_feedback_forward_relative_error": 1.1545552064072002,
        "min_feedback_forward_cosine": 0.10470203310251236
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 782,
      "train_examples": 50000,
      "train_loss": 1.6967413638687134
    },
    {
      "epoch": 3,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3334531784057617,
        "mean_feedback_forward_cosine": 0.3631359176202254,
        "mean_feedback_forward_relative_error": 1.1180878097360785,
        "min_feedback_forward_cosine": 0.11801935732364655
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1173,
      "train_examples": 50000,
      "train_loss": 1.4307723350524901
    },
    {
      "epoch": 4,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3191616535186768,
        "mean_feedback_forward_cosine": 0.41192043640396814,
        "mean_feedback_forward_relative_error": 1.0758537660945546,
        "min_feedback_forward_cosine": 0.13609740138053894
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1564,
      "train_examples": 50000,
      "train_loss": 1.1842613387298584
    },
    {
      "epoch": 5,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3043830394744873,
        "mean_feedback_forward_cosine": 0.4591585630720312,
        "mean_feedback_forward_relative_error": 1.0326683066108011,
        "min_feedback_forward_cosine": 0.15394139289855957
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1955,
      "train_examples": 50000,
      "train_loss": 0.9762814709091187
    },
    {
      "epoch": 6,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2937240600585938,
        "mean_feedback_forward_cosine": 0.4997470823201266,
        "mean_feedback_forward_relative_error": 0.9934693618254228,
        "min_feedback_forward_cosine": 0.1700705885887146
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2346,
      "train_examples": 50000,
      "train_loss": 0.8191733999633789
    },
    {
      "epoch": 7,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2820374965667725,
        "mean_feedback_forward_cosine": 0.5342417137189345,
        "mean_feedback_forward_relative_error": 0.9584944692524997,
        "min_feedback_forward_cosine": 0.1848730593919754
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2737,
      "train_examples": 50000,
      "train_loss": 0.7156849175262451
    },
    {
      "epoch": 8,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.270569920539856,
        "mean_feedback_forward_cosine": 0.5640160666270689,
        "mean_feedback_forward_relative_error": 0.9268558773127469,
        "min_feedback_forward_cosine": 0.19986508786678314
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3128,
      "train_examples": 50000,
      "train_loss": 0.6479228650474549
    },
    {
      "epoch": 9,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.257145881652832,
        "mean_feedback_forward_cosine": 0.5917187715118581,
        "mean_feedback_forward_relative_error": 0.8961032244292173,
        "min_feedback_forward_cosine": 0.2162085771560669
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3519,
      "train_examples": 50000,
      "train_loss": 0.5960064146232605
    },
    {
      "epoch": 10,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2436530590057373,
        "mean_feedback_forward_cosine": 0.6161474466323853,
        "mean_feedback_forward_relative_error": 0.867996008829637,
        "min_feedback_forward_cosine": 0.23284927010536194
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3910,
      "train_examples": 50000,
      "train_loss": 0.5511851186561585
    },
    {
      "epoch": 11,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.228519082069397,
        "mean_feedback_forward_cosine": 0.6386087173765356,
        "mean_feedback_forward_relative_error": 0.8411842422051863,
        "min_feedback_forward_cosine": 0.25119492411613464
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4301,
      "train_examples": 50000,
      "train_loss": 0.5161278319740296
    },
    {
      "epoch": 12,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2122453451156616,
        "mean_feedback_forward_cosine": 0.659218141707507,
        "mean_feedback_forward_relative_error": 0.8157952037724582,
        "min_feedback_forward_cosine": 0.2712697386741638
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4692,
      "train_examples": 50000,
      "train_loss": 0.48363601528167727
    },
    {
      "epoch": 13,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1960726976394653,
        "mean_feedback_forward_cosine": 0.6787421963431618,
        "mean_feedback_forward_relative_error": 0.7907431545582685,
        "min_feedback_forward_cosine": 0.2901216745376587
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5083,
      "train_examples": 50000,
      "train_loss": 0.4618261950588226
    },
    {
      "epoch": 14,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1796756982803345,
        "mean_feedback_forward_cosine": 0.6973863617940382,
        "mean_feedback_forward_relative_error": 0.7662305073304609,
        "min_feedback_forward_cosine": 0.31009507179260254
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5474,
      "train_examples": 50000,
      "train_loss": 0.4427673635292053
    },
    {
      "epoch": 15,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.165794014930725,
        "mean_feedback_forward_cosine": 0.7144045293331146,
        "mean_feedback_forward_relative_error": 0.7431726366281509,
        "min_feedback_forward_cosine": 0.3328298032283783
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5865,
      "train_examples": 50000,
      "train_loss": 0.41954453619003296
    },
    {
      "epoch": 16,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1517013311386108,
        "mean_feedback_forward_cosine": 0.7306086350571025,
        "mean_feedback_forward_relative_error": 0.7205345245924863,
        "min_feedback_forward_cosine": 0.3534374535083771
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 6256,
      "train_examples": 50000,
      "train_loss": 0.4039508693885803
    },
    {
      "epoch": 17,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1365071535110474,
        "mean_feedback_forward_cosine": 0.7460741254416379,
        "mean_feedback_forward_relative_error": 0.6985243339430202,
        "min_feedback_forward_cosine": 0.3746267855167389
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 6647,
      "train_examples": 50000,
      "train_loss": 0.3821706154251099
    },
    {
      "epoch": 18,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1201761960983276,
        "mean_feedback_forward_cosine": 0.7606539937582883,
        "mean_feedback_forward_relative_error": 0.6771352933211761,
        "min_feedback_forward_cosine": 0.3938027620315552
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7038,
      "train_examples": 50000,
      "train_loss": 0.3685064894771576
    },
    {
      "epoch": 19,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1034852266311646,
        "mean_feedback_forward_cosine": 0.7745427044955167,
        "mean_feedback_forward_relative_error": 0.6560068325562911,
        "min_feedback_forward_cosine": 0.4126436710357666
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7429,
      "train_examples": 50000,
      "train_loss": 0.3599610644435883
    },
    {
      "epoch": 20,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0848698616027832,
        "mean_feedback_forward_cosine": 0.7881048738956451,
        "mean_feedback_forward_relative_error": 0.6349188978021795,
        "min_feedback_forward_cosine": 0.433307945728302
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7820,
      "train_examples": 50000,
      "train_loss": 0.3474660382270813
    },
    {
      "epoch": 21,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0681661367416382,
        "mean_feedback_forward_cosine": 0.800181284275922,
        "mean_feedback_forward_relative_error": 0.6155560723759911,
        "min_feedback_forward_cosine": 0.45181435346603394
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8211,
      "train_examples": 50000,
      "train_loss": 0.3313999942684174
    },
    {
      "epoch": 22,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.04939603805542,
        "mean_feedback_forward_cosine": 0.8119089007377625,
        "mean_feedback_forward_relative_error": 0.5961991448294033,
        "min_feedback_forward_cosine": 0.47210100293159485
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8602,
      "train_examples": 50000,
      "train_loss": 0.32168210739135744
    },
    {
      "epoch": 23,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0293253660202026,
        "mean_feedback_forward_cosine": 0.8231192323294553,
        "mean_feedback_forward_relative_error": 0.577244947715239,
        "min_feedback_forward_cosine": 0.4929039776325226
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8993,
      "train_examples": 50000,
      "train_loss": 0.31462405000686644
    },
    {
      "epoch": 24,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0082814693450928,
        "mean_feedback_forward_cosine": 0.8338727604259144,
        "mean_feedback_forward_relative_error": 0.558591572263024,
        "min_feedback_forward_cosine": 0.5139169692993164
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9384,
      "train_examples": 50000,
      "train_loss": 0.3024618870449066
    },
    {
      "epoch": 25,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.987269401550293,
        "mean_feedback_forward_cosine": 0.8439520521597429,
        "mean_feedback_forward_relative_error": 0.54047522761605,
        "min_feedback_forward_cosine": 0.534113883972168
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9775,
      "train_examples": 50000,
      "train_loss": 0.29681352372169495
    },
    {
      "epoch": 26,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9666692018508911,
        "mean_feedback_forward_cosine": 0.8535854989832098,
        "mean_feedback_forward_relative_error": 0.5227361795577136,
        "min_feedback_forward_cosine": 0.5536783933639526
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10166,
      "train_examples": 50000,
      "train_loss": 0.291030342874527
    },
    {
      "epoch": 27,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.946130096912384,
        "mean_feedback_forward_cosine": 0.8625394571911205,
        "mean_feedback_forward_relative_error": 0.5056248185309496,
        "min_feedback_forward_cosine": 0.5723249912261963
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10557,
      "train_examples": 50000,
      "train_loss": 0.2811413796710968
    },
    {
      "epoch": 28,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9224547743797302,
        "mean_feedback_forward_cosine": 0.8712533885782415,
        "mean_feedback_forward_relative_error": 0.48866380696946926,
        "min_feedback_forward_cosine": 0.5937190055847168
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10948,
      "train_examples": 50000,
      "train_loss": 0.27247773029327393
    },
    {
      "epoch": 29,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8997247815132141,
        "mean_feedback_forward_cosine": 0.87938373522325,
        "mean_feedback_forward_relative_error": 0.4724398217417977,
        "min_feedback_forward_cosine": 0.6140031218528748
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11339,
      "train_examples": 50000,
      "train_loss": 0.26573392446517946
    },
    {
      "epoch": 30,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8767585158348083,
        "mean_feedback_forward_cosine": 0.8873388496312228,
        "mean_feedback_forward_relative_error": 0.45595376071604815,
        "min_feedback_forward_cosine": 0.6334506869316101
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11730,
      "train_examples": 50000,
      "train_loss": 0.2651367131471634
    },
    {
      "epoch": 31,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8535721898078918,
        "mean_feedback_forward_cosine": 0.8947497736323964,
        "mean_feedback_forward_relative_error": 0.4401968778534369,
        "min_feedback_forward_cosine": 0.652430534362793
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12121,
      "train_examples": 50000,
      "train_loss": 0.261038997964859
    },
    {
      "epoch": 32,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8301363587379456,
        "mean_feedback_forward_cosine": 0.9016447034749118,
        "mean_feedback_forward_relative_error": 0.4251384423537688,
        "min_feedback_forward_cosine": 0.6714699268341064
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12512,
      "train_examples": 50000,
      "train_loss": 0.24747045076847077
    },
    {
      "epoch": 33,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8073676228523254,
        "mean_feedback_forward_cosine": 0.9081655393947254,
        "mean_feedback_forward_relative_error": 0.41030791076746853,
        "min_feedback_forward_cosine": 0.6888424158096313
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12903,
      "train_examples": 50000,
      "train_loss": 0.24297086081027985
    },
    {
      "epoch": 34,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7856544256210327,
        "mean_feedback_forward_cosine": 0.9142789288000627,
        "mean_feedback_forward_relative_error": 0.39597937315702436,
        "min_feedback_forward_cosine": 0.7051998972892761
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13294,
      "train_examples": 50000,
      "train_loss": 0.24031951283931732
    },
    {
      "epoch": 35,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7637181878089905,
        "mean_feedback_forward_cosine": 0.9201499971476468,
        "mean_feedback_forward_relative_error": 0.3818268209695816,
        "min_feedback_forward_cosine": 0.7213693857192993
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13685,
      "train_examples": 50000,
      "train_loss": 0.23635845373630524
    },
    {
      "epoch": 36,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7411061525344849,
        "mean_feedback_forward_cosine": 0.9256504665721547,
        "mean_feedback_forward_relative_error": 0.36806188930164685,
        "min_feedback_forward_cosine": 0.7375943660736084
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14076,
      "train_examples": 50000,
      "train_loss": 0.23369537858963013
    },
    {
      "epoch": 37,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7196806073188782,
        "mean_feedback_forward_cosine": 0.9307226386937228,
        "mean_feedback_forward_relative_error": 0.3549185817891901,
        "min_feedback_forward_cosine": 0.7523044943809509
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14467,
      "train_examples": 50000,
      "train_loss": 0.2313125407028198
    },
    {
      "epoch": 38,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6974388957023621,
        "mean_feedback_forward_cosine": 0.9354210929437117,
        "mean_feedback_forward_relative_error": 0.342401736703786,
        "min_feedback_forward_cosine": 0.7674288749694824
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14858,
      "train_examples": 50000,
      "train_loss": 0.22255386281967163
    },
    {
      "epoch": 39,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6743432283401489,
        "mean_feedback_forward_cosine": 0.939997334913774,
        "mean_feedback_forward_relative_error": 0.32986792014403776,
        "min_feedback_forward_cosine": 0.7824465036392212
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15249,
      "train_examples": 50000,
      "train_loss": 0.223271989402771
    },
    {
      "epoch": 40,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6519086956977844,
        "mean_feedback_forward_cosine": 0.9441849589347839,
        "mean_feedback_forward_relative_error": 0.3179208685051311,
        "min_feedback_forward_cosine": 0.7965607643127441
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15640,
      "train_examples": 50000,
      "train_loss": 0.21751073798656465
    },
    {
      "epoch": 41,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6300700306892395,
        "mean_feedback_forward_cosine": 0.9482373237609864,
        "mean_feedback_forward_relative_error": 0.3059993200681426,
        "min_feedback_forward_cosine": 0.809735119342804
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16031,
      "train_examples": 50000,
      "train_loss": 0.21728030968666076
    },
    {
      "epoch": 42,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6099371910095215,
        "mean_feedback_forward_cosine": 0.9519010370427912,
        "mean_feedback_forward_relative_error": 0.29486370913007043,
        "min_feedback_forward_cosine": 0.8216880559921265
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16422,
      "train_examples": 50000,
      "train_loss": 0.20883438779830932
    },
    {
      "epoch": 43,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5901137590408325,
        "mean_feedback_forward_cosine": 0.9552752581509677,
        "mean_feedback_forward_relative_error": 0.2841600298881531,
        "min_feedback_forward_cosine": 0.8328014612197876
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16813,
      "train_examples": 50000,
      "train_loss": 0.20926959754467012
    },
    {
      "epoch": 44,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5698909759521484,
        "mean_feedback_forward_cosine": 0.9585013281215321,
        "mean_feedback_forward_relative_error": 0.273653789406473,
        "min_feedback_forward_cosine": 0.844026505947113
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17204,
      "train_examples": 50000,
      "train_loss": 0.20790495951652527
    },
    {
      "epoch": 45,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5506214499473572,
        "mean_feedback_forward_cosine": 0.9614172577857971,
        "mean_feedback_forward_relative_error": 0.26380565762519836,
        "min_feedback_forward_cosine": 0.8544111847877502
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17595,
      "train_examples": 50000,
      "train_loss": 0.2017436584854126
    },
    {
      "epoch": 46,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5307701230049133,
        "mean_feedback_forward_cosine": 0.9642623077739368,
        "mean_feedback_forward_relative_error": 0.25384207489815624,
        "min_feedback_forward_cosine": 0.8647032976150513
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17986,
      "train_examples": 50000,
      "train_loss": 0.19571260312080382
    },
    {
      "epoch": 47,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5120128393173218,
        "mean_feedback_forward_cosine": 0.9669614347544584,
        "mean_feedback_forward_relative_error": 0.2439642841165716,
        "min_feedback_forward_cosine": 0.8739529848098755
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 18377,
      "train_examples": 50000,
      "train_loss": 0.1975364964056015
    },
    {
      "epoch": 48,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.4943452775478363,
        "mean_feedback_forward_cosine": 0.9693642713806846,
        "mean_feedback_forward_relative_error": 0.23483136296272278,
        "min_feedback_forward_cosine": 0.8824831247329712
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 18768,
      "train_examples": 50000,
      "train_loss": 0.19639045250177384
    },
    {
      "epoch": 49,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.4769802987575531,
        "mean_feedback_forward_cosine": 0.9715972369367426,
        "mean_feedback_forward_relative_error": 0.22603400234471668,
        "min_feedback_forward_cosine": 0.8905482292175293
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19159,
      "train_examples": 50000,
      "train_loss": 0.18923484493255616
    },
    {
      "epoch": 50,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.46077466011047363,
        "mean_feedback_forward_cosine": 0.9736243302171881,
        "mean_feedback_forward_relative_error": 0.21773624020543966,
        "min_feedback_forward_cosine": 0.8977358341217041
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19550,
      "train_examples": 50000,
      "train_loss": 0.18611038329124452
    },
    {
      "epoch": 51,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.44443535804748535,
        "mean_feedback_forward_cosine": 0.9755336100404913,
        "mean_feedback_forward_relative_error": 0.2096899394284595,
        "min_feedback_forward_cosine": 0.904762864112854
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19941,
      "train_examples": 50000,
      "train_loss": 0.18421099987983702
    },
    {
      "epoch": 52,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.42707616090774536,
        "mean_feedback_forward_cosine": 0.9773360046473416,
        "mean_feedback_forward_relative_error": 0.20185066983103753,
        "min_feedback_forward_cosine": 0.9120082855224609
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20332,
      "train_examples": 50000,
      "train_loss": 0.18032447082519532
    },
    {
      "epoch": 53,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.41098570823669434,
        "mean_feedback_forward_cosine": 0.9790595271370628,
        "mean_feedback_forward_relative_error": 0.1940204645422372,
        "min_feedback_forward_cosine": 0.9184935092926025
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20723,
      "train_examples": 50000,
      "train_loss": 0.1868881555366516
    },
    {
      "epoch": 54,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3946766257286072,
        "mean_feedback_forward_cosine": 0.9806340380148454,
        "mean_feedback_forward_relative_error": 0.18661480302160435,
        "min_feedback_forward_cosine": 0.9246987104415894
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21114,
      "train_examples": 50000,
      "train_loss": 0.17931044955730438
    },
    {
      "epoch": 55,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.37967368960380554,
        "mean_feedback_forward_cosine": 0.9820866823196411,
        "mean_feedback_forward_relative_error": 0.17948700020259076,
        "min_feedback_forward_cosine": 0.9302462339401245
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21505,
      "train_examples": 50000,
      "train_loss": 0.1762636404657364
    },
    {
      "epoch": 56,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.36562249064445496,
        "mean_feedback_forward_cosine": 0.9834293506362222,
        "mean_feedback_forward_relative_error": 0.17262107696045528,
        "min_feedback_forward_cosine": 0.9352699518203735
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21896,
      "train_examples": 50000,
      "train_loss": 0.1766922233390808
    },
    {
      "epoch": 57,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.35152167081832886,
        "mean_feedback_forward_cosine": 0.9846597454764626,
        "mean_feedback_forward_relative_error": 0.16610525501045315,
        "min_feedback_forward_cosine": 0.9401300549507141
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22287,
      "train_examples": 50000,
      "train_loss": 0.17598136737823486
    },
    {
      "epoch": 58,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3375950753688812,
        "mean_feedback_forward_cosine": 0.9858143698085439,
        "mean_feedback_forward_relative_error": 0.15978123186664148,
        "min_feedback_forward_cosine": 0.9447086453437805
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22678,
      "train_examples": 50000,
      "train_loss": 0.17427285524368286
    },
    {
      "epoch": 59,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3254486322402954,
        "mean_feedback_forward_cosine": 0.9868530143391002,
        "mean_feedback_forward_relative_error": 0.15380413701588458,
        "min_feedback_forward_cosine": 0.94859379529953
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23069,
      "train_examples": 50000,
      "train_loss": 0.16943655662059784
    },
    {
      "epoch": 60,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3130016028881073,
        "mean_feedback_forward_cosine": 0.9878478624603965,
        "mean_feedback_forward_relative_error": 0.14787317527966065,
        "min_feedback_forward_cosine": 0.9523912668228149
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23460,
      "train_examples": 50000,
      "train_loss": 0.17375968336105346
    },
    {
      "epoch": 61,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3006737530231476,
        "mean_feedback_forward_cosine": 0.9887408408251676,
        "mean_feedback_forward_relative_error": 0.14235474067655476,
        "min_feedback_forward_cosine": 0.9560116529464722
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23851,
      "train_examples": 50000,
      "train_loss": 0.1664853154194355
    },
    {
      "epoch": 62,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2892468273639679,
        "mean_feedback_forward_cosine": 0.9895924654873934,
        "mean_feedback_forward_relative_error": 0.13685577057979323,
        "min_feedback_forward_cosine": 0.9592477083206177
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24242,
      "train_examples": 50000,
      "train_loss": 0.16777808388233184
    },
    {
      "epoch": 63,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2781123220920563,
        "mean_feedback_forward_cosine": 0.9903776082125577,
        "mean_feedback_forward_relative_error": 0.13159775713627989,
        "min_feedback_forward_cosine": 0.9622933864593506
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24633,
      "train_examples": 50000,
      "train_loss": 0.16732357576370238
    },
    {
      "epoch": 64,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2673693597316742,
        "mean_feedback_forward_cosine": 0.9910991148515181,
        "mean_feedback_forward_relative_error": 0.12656967727975413,
        "min_feedback_forward_cosine": 0.9651110172271729
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25024,
      "train_examples": 50000,
      "train_loss": 0.1668574471092224
    },
    {
      "epoch": 65,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2565230429172516,
        "mean_feedback_forward_cosine": 0.9917782989415256,
        "mean_feedback_forward_relative_error": 0.12167377007955854,
        "min_feedback_forward_cosine": 0.9678463339805603
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25415,
      "train_examples": 50000,
      "train_loss": 0.1682041015291214
    },
    {
      "epoch": 66,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.246418297290802,
        "mean_feedback_forward_cosine": 0.9923990683122115,
        "mean_feedback_forward_relative_error": 0.11701159981841391,
        "min_feedback_forward_cosine": 0.9703002572059631
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25806,
      "train_examples": 50000,
      "train_loss": 0.16243797614574432
    },
    {
      "epoch": 67,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.23698997497558594,
        "mean_feedback_forward_cosine": 0.9929798906499689,
        "mean_feedback_forward_relative_error": 0.11245247261090711,
        "min_feedback_forward_cosine": 0.9725117683410645
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26197,
      "train_examples": 50000,
      "train_loss": 0.16206168502807616
    },
    {
      "epoch": 68,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.22819016873836517,
        "mean_feedback_forward_cosine": 0.9934903372417797,
        "mean_feedback_forward_relative_error": 0.10829746635122732,
        "min_feedback_forward_cosine": 0.9744926691055298
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26588,
      "train_examples": 50000,
      "train_loss": 0.1601631025314331
    },
    {
      "epoch": 69,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2193627506494522,
        "mean_feedback_forward_cosine": 0.9939808195287531,
        "mean_feedback_forward_relative_error": 0.10415428670292551,
        "min_feedback_forward_cosine": 0.9764065742492676
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26979,
      "train_examples": 50000,
      "train_loss": 0.15879637899398805
    },
    {
      "epoch": 70,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.21048295497894287,
        "mean_feedback_forward_cosine": 0.9944382244890386,
        "mean_feedback_forward_relative_error": 0.10016306428746743,
        "min_feedback_forward_cosine": 0.9782607555389404
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27370,
      "train_examples": 50000,
      "train_loss": 0.160089029340744
    },
    {
      "epoch": 71,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.20216532051563263,
        "mean_feedback_forward_cosine": 0.9948591644113715,
        "mean_feedback_forward_relative_error": 0.0963318647308783,
        "min_feedback_forward_cosine": 0.9799298644065857
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27761,
      "train_examples": 50000,
      "train_loss": 0.15733842891693114
    },
    {
      "epoch": 72,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.19413991272449493,
        "mean_feedback_forward_cosine": 0.995243741165508,
        "mean_feedback_forward_relative_error": 0.09266701493073594,
        "min_feedback_forward_cosine": 0.9814789891242981
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28152,
      "train_examples": 50000,
      "train_loss": 0.15482519105911255
    },
    {
      "epoch": 73,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.18668678402900696,
        "mean_feedback_forward_cosine": 0.9955994064157659,
        "mean_feedback_forward_relative_error": 0.08915054053068161,
        "min_feedback_forward_cosine": 0.9828635454177856
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28543,
      "train_examples": 50000,
      "train_loss": 0.1533050349807739
    },
    {
      "epoch": 74,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.17943738400936127,
        "mean_feedback_forward_cosine": 0.9959320068359375,
        "mean_feedback_forward_relative_error": 0.08572588434273547,
        "min_feedback_forward_cosine": 0.9841609001159668
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28934,
      "train_examples": 50000,
      "train_loss": 0.15046897461891173
    },
    {
      "epoch": 75,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.17239177227020264,
        "mean_feedback_forward_cosine": 0.99624718319286,
        "mean_feedback_forward_relative_error": 0.08235500065440482,
        "min_feedback_forward_cosine": 0.9853790402412415
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29325,
      "train_examples": 50000,
      "train_loss": 0.15597980010747908
    },
    {
      "epoch": 76,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.16524744033813477,
        "mean_feedback_forward_cosine": 0.9965404543009672,
        "mean_feedback_forward_relative_error": 0.07910626672885634,
        "min_feedback_forward_cosine": 0.9865565299987793
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29716,
      "train_examples": 50000,
      "train_loss": 0.15417349669456482
    },
    {
      "epoch": 77,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.15881438553333282,
        "mean_feedback_forward_cosine": 0.9967974153431979,
        "mean_feedback_forward_relative_error": 0.07613672224635427,
        "min_feedback_forward_cosine": 0.9875755906105042
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30107,
      "train_examples": 50000,
      "train_loss": 0.1473866323184967
    },
    {
      "epoch": 78,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.15201766788959503,
        "mean_feedback_forward_cosine": 0.9970540989529003,
        "mean_feedback_forward_relative_error": 0.0730514292689887,
        "min_feedback_forward_cosine": 0.988610029220581
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30498,
      "train_examples": 50000,
      "train_loss": 0.15314912130355834
    },
    {
      "epoch": 79,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.14590775966644287,
        "mean_feedback_forward_cosine": 0.9972771482034163,
        "mean_feedback_forward_relative_error": 0.07023630430075255,
        "min_feedback_forward_cosine": 0.9895028471946716
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30889,
      "train_examples": 50000,
      "train_loss": 0.15089951363563536
    },
    {
      "epoch": 80,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.13983440399169922,
        "mean_feedback_forward_cosine": 0.9974843111905185,
        "mean_feedback_forward_relative_error": 0.06753264333714139,
        "min_feedback_forward_cosine": 0.9903545379638672
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 31280,
      "train_examples": 50000,
      "train_loss": 0.15078624309539795
    },
    {
      "epoch": 81,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.13432933390140533,
        "mean_feedback_forward_cosine": 0.9976726553656838,
        "mean_feedback_forward_relative_error": 0.06496261497112837,
        "min_feedback_forward_cosine": 0.9910942316055298
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 31671,
      "train_examples": 50000,
      "train_loss": 0.14839280843496322
    },
    {
      "epoch": 82,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1289684921503067,
        "mean_feedback_forward_cosine": 0.9978530916300687,
        "mean_feedback_forward_relative_error": 0.062406730838119984,
        "min_feedback_forward_cosine": 0.9917852878570557
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32062,
      "train_examples": 50000,
      "train_loss": 0.14823583701610565
    },
    {
      "epoch": 83,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.12417666614055634,
        "mean_feedback_forward_cosine": 0.9980134508826516,
        "mean_feedback_forward_relative_error": 0.06003500163893808,
        "min_feedback_forward_cosine": 0.9923818111419678
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32453,
      "train_examples": 50000,
      "train_loss": 0.14464116917133332
    },
    {
      "epoch": 84,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.11920725554227829,
        "mean_feedback_forward_cosine": 0.9981636946851556,
        "mean_feedback_forward_relative_error": 0.05773626547306776,
        "min_feedback_forward_cosine": 0.9929776191711426
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32844,
      "train_examples": 50000,
      "train_loss": 0.14492770351409912
    },
    {
      "epoch": 85,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.11448901146650314,
        "mean_feedback_forward_cosine": 0.99830322265625,
        "mean_feedback_forward_relative_error": 0.05551600430838086,
        "min_feedback_forward_cosine": 0.9935189485549927
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33235,
      "train_examples": 50000,
      "train_loss": 0.14268309606552124
    },
    {
      "epoch": 86,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.10983197391033173,
        "mean_feedback_forward_cosine": 0.9984360218048096,
        "mean_feedback_forward_relative_error": 0.053309309702705254,
        "min_feedback_forward_cosine": 0.9940335750579834
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33626,
      "train_examples": 50000,
      "train_loss": 0.1478362784576416
    },
    {
      "epoch": 87,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.10562899708747864,
        "mean_feedback_forward_cosine": 0.9985541972247037,
        "mean_feedback_forward_relative_error": 0.05125798300247301,
        "min_feedback_forward_cosine": 0.9944791793823242
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34017,
      "train_examples": 50000,
      "train_loss": 0.1437551615524292
    },
    {
      "epoch": 88,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.10163402557373047,
        "mean_feedback_forward_cosine": 0.9986639445478266,
        "mean_feedback_forward_relative_error": 0.04928394603458318,
        "min_feedback_forward_cosine": 0.9948862791061401
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34408,
      "train_examples": 50000,
      "train_loss": 0.14351468017578126
    },
    {
      "epoch": 89,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.09729874134063721,
        "mean_feedback_forward_cosine": 0.9987684997645292,
        "mean_feedback_forward_relative_error": 0.047331926243549045,
        "min_feedback_forward_cosine": 0.9953108429908752
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34799,
      "train_examples": 50000,
      "train_loss": 0.14870515768051148
    },
    {
      "epoch": 90,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.09344491362571716,
        "mean_feedback_forward_cosine": 0.9988601771267978,
        "mean_feedback_forward_relative_error": 0.04554461816495115,
        "min_feedback_forward_cosine": 0.9956722259521484
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35190,
      "train_examples": 50000,
      "train_loss": 0.14165220482349397
    },
    {
      "epoch": 91,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.08984030038118362,
        "mean_feedback_forward_cosine": 0.9989478772336786,
        "mean_feedback_forward_relative_error": 0.043766251460395075,
        "min_feedback_forward_cosine": 0.995998740196228
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35581,
      "train_examples": 50000,
      "train_loss": 0.1420630505466461
    },
    {
      "epoch": 92,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.08653393387794495,
        "mean_feedback_forward_cosine": 0.9990258845415982,
        "mean_feedback_forward_relative_error": 0.04212138226086443,
        "min_feedback_forward_cosine": 0.9962879419326782
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35972,
      "train_examples": 50000,
      "train_loss": 0.13872229014873505
    },
    {
      "epoch": 93,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.08304337412118912,
        "mean_feedback_forward_cosine": 0.9991016897288236,
        "mean_feedback_forward_relative_error": 0.04045714531093836,
        "min_feedback_forward_cosine": 0.9965806603431702
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 36363,
      "train_examples": 50000,
      "train_loss": 0.1445293878555298
    },
    {
      "epoch": 94,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.07964146882295609,
        "mean_feedback_forward_cosine": 0.9991688966751099,
        "mean_feedback_forward_relative_error": 0.03892969995398413,
        "min_feedback_forward_cosine": 0.9968552589416504
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 36754,
      "train_examples": 50000,
      "train_loss": 0.13843780771970748
    },
    {
      "epoch": 95,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.07634511590003967,
        "mean_feedback_forward_cosine": 0.9992327451705932,
        "mean_feedback_forward_relative_error": 0.03741777705197984,
        "min_feedback_forward_cosine": 0.9971089363098145
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 37145,
      "train_examples": 50000,
      "train_loss": 0.13654371682167052
    },
    {
      "epoch": 96,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.07344254851341248,
        "mean_feedback_forward_cosine": 0.9992919184944846,
        "mean_feedback_forward_relative_error": 0.03594837166707624,
        "min_feedback_forward_cosine": 0.9973240494728088
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 37536,
      "train_examples": 50000,
      "train_loss": 0.1398654334306717
    },
    {
      "epoch": 97,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.07045339047908783,
        "mean_feedback_forward_cosine": 0.9993484518744729,
        "mean_feedback_forward_relative_error": 0.03449128752743656,
        "min_feedback_forward_cosine": 0.9975365996360779
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 37927,
      "train_examples": 50000,
      "train_loss": 0.14294681231498718
    },
    {
      "epoch": 98,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06773552298545837,
        "mean_feedback_forward_cosine": 0.9993969202041626,
        "mean_feedback_forward_relative_error": 0.033191636018455026,
        "min_feedback_forward_cosine": 0.9977228045463562
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 38318,
      "train_examples": 50000,
      "train_loss": 0.13541974694252015
    },
    {
      "epoch": 99,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06507350504398346,
        "mean_feedback_forward_cosine": 0.9994427128271623,
        "mean_feedback_forward_relative_error": 0.031913321566852657,
        "min_feedback_forward_cosine": 0.9978986978530884
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 38709,
      "train_examples": 50000,
      "train_loss": 0.14026688925266265
    },
    {
      "epoch": 100,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0627213791012764,
        "mean_feedback_forward_cosine": 0.9994821851903742,
        "mean_feedback_forward_relative_error": 0.030764484193853355,
        "min_feedback_forward_cosine": 0.9980470538139343
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 39100,
      "train_examples": 50000,
      "train_loss": 0.13205082747459412
    },
    {
      "epoch": 101,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06269843876361847,
        "mean_feedback_forward_cosine": 0.9994824994694103,
        "mean_feedback_forward_relative_error": 0.03075404701415788,
        "min_feedback_forward_cosine": 0.9980485439300537
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39491,
      "train_examples": 50000,
      "train_loss": 0.06759102570533752
    },
    {
      "epoch": 102,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06268610805273056,
        "mean_feedback_forward_cosine": 0.9994826923717152,
        "mean_feedback_forward_relative_error": 0.03074822404675863,
        "min_feedback_forward_cosine": 0.9980494976043701
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39882,
      "train_examples": 50000,
      "train_loss": 0.045256944432854655
    },
    {
      "epoch": 103,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06267379224300385,
        "mean_feedback_forward_cosine": 0.9994828408414668,
        "mean_feedback_forward_relative_error": 0.03074289435012774,
        "min_feedback_forward_cosine": 0.9980503916740417
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40273,
      "train_examples": 50000,
      "train_loss": 0.04034666366100311
    },
    {
      "epoch": 104,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06266102194786072,
        "mean_feedback_forward_cosine": 0.9994830001484264,
        "mean_feedback_forward_relative_error": 0.030737932263450188,
        "min_feedback_forward_cosine": 0.9980511665344238
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40664,
      "train_examples": 50000,
      "train_loss": 0.03293282743215561
    },
    {
      "epoch": 105,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06264709681272507,
        "mean_feedback_forward_cosine": 0.9994831702925943,
        "mean_feedback_forward_relative_error": 0.030732802640308032,
        "min_feedback_forward_cosine": 0.9980521202087402
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41055,
      "train_examples": 50000,
      "train_loss": 0.03022145635008812
    },
    {
      "epoch": 106,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06263425946235657,
        "mean_feedback_forward_cosine": 0.9994833133437416,
        "mean_feedback_forward_relative_error": 0.030728215766562657,
        "min_feedback_forward_cosine": 0.9980529546737671
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41446,
      "train_examples": 50000,
      "train_loss": 0.025835701021552086
    },
    {
      "epoch": 107,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06262177228927612,
        "mean_feedback_forward_cosine": 0.999483428218148,
        "mean_feedback_forward_relative_error": 0.030723788335242055,
        "min_feedback_forward_cosine": 0.9980536699295044
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41837,
      "train_examples": 50000,
      "train_loss": 0.02356987555027008
    },
    {
      "epoch": 108,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06260990351438522,
        "mean_feedback_forward_cosine": 0.9994835582646456,
        "mean_feedback_forward_relative_error": 0.03071964384818619,
        "min_feedback_forward_cosine": 0.998054563999176
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42228,
      "train_examples": 50000,
      "train_loss": 0.021534102993607522
    },
    {
      "epoch": 109,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06259550154209137,
        "mean_feedback_forward_cosine": 0.9994837110692805,
        "mean_feedback_forward_relative_error": 0.03071519649502906,
        "min_feedback_forward_cosine": 0.9980553388595581
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42619,
      "train_examples": 50000,
      "train_loss": 0.020321601420640944
    },
    {
      "epoch": 110,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0625816360116005,
        "mean_feedback_forward_cosine": 0.9994838573715904,
        "mean_feedback_forward_relative_error": 0.03071072819558057,
        "min_feedback_forward_cosine": 0.9980563521385193
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43010,
      "train_examples": 50000,
      "train_loss": 0.019604212568700312
    },
    {
      "epoch": 111,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06256827712059021,
        "mean_feedback_forward_cosine": 0.9994839744134383,
        "mean_feedback_forward_relative_error": 0.03070665564049374,
        "min_feedback_forward_cosine": 0.9980570673942566
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43401,
      "train_examples": 50000,
      "train_loss": 0.016796411416381597
    },
    {
      "epoch": 112,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06255496293306351,
        "mean_feedback_forward_cosine": 0.9994841022924943,
        "mean_feedback_forward_relative_error": 0.03070269122042439,
        "min_feedback_forward_cosine": 0.9980579614639282
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43792,
      "train_examples": 50000,
      "train_loss": 0.015727011843919753
    },
    {
      "epoch": 113,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06253934651613235,
        "mean_feedback_forward_cosine": 0.9994842648506165,
        "mean_feedback_forward_relative_error": 0.030698163578794763,
        "min_feedback_forward_cosine": 0.998059093952179
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44183,
      "train_examples": 50000,
      "train_loss": 0.015990155968666077
    },
    {
      "epoch": 114,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06252351403236389,
        "mean_feedback_forward_cosine": 0.9994844100692055,
        "mean_feedback_forward_relative_error": 0.03069376213984056,
        "min_feedback_forward_cosine": 0.9980600476264954
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44574,
      "train_examples": 50000,
      "train_loss": 0.01549170794636011
    },
    {
      "epoch": 115,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06250815838575363,
        "mean_feedback_forward_cosine": 0.9994845433668657,
        "mean_feedback_forward_relative_error": 0.030689492339098994,
        "min_feedback_forward_cosine": 0.9980608820915222
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44965,
      "train_examples": 50000,
      "train_loss": 0.014437230432629585
    },
    {
      "epoch": 116,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06249437481164932,
        "mean_feedback_forward_cosine": 0.9994846593249928,
        "mean_feedback_forward_relative_error": 0.030685534599152478,
        "min_feedback_forward_cosine": 0.9980616569519043
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45356,
      "train_examples": 50000,
      "train_loss": 0.01300610273450613
    },
    {
      "epoch": 117,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062480244785547256,
        "mean_feedback_forward_cosine": 0.9994848132133484,
        "mean_feedback_forward_relative_error": 0.03068151062523777,
        "min_feedback_forward_cosine": 0.9980626106262207
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45747,
      "train_examples": 50000,
      "train_loss": 0.01203647702410817
    },
    {
      "epoch": 118,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0624653697013855,
        "mean_feedback_forward_cosine": 0.9994849421761253,
        "mean_feedback_forward_relative_error": 0.030677358492870222,
        "min_feedback_forward_cosine": 0.9980636239051819
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46138,
      "train_examples": 50000,
      "train_loss": 0.012282160293906927
    },
    {
      "epoch": 119,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06244941055774689,
        "mean_feedback_forward_cosine": 0.9994850819761103,
        "mean_feedback_forward_relative_error": 0.030673144935545595,
        "min_feedback_forward_cosine": 0.9980645775794983
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46529,
      "train_examples": 50000,
      "train_loss": 0.011743366521783174
    },
    {
      "epoch": 120,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062433239072561264,
        "mean_feedback_forward_cosine": 0.9994852185249329,
        "mean_feedback_forward_relative_error": 0.03066888998516581,
        "min_feedback_forward_cosine": 0.9980655312538147
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46920,
      "train_examples": 50000,
      "train_loss": 0.011445523285418749
    },
    {
      "epoch": 121,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06241831183433533,
        "mean_feedback_forward_cosine": 0.9994853594086387,
        "mean_feedback_forward_relative_error": 0.03066485823047432,
        "min_feedback_forward_cosine": 0.9980664849281311
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47311,
      "train_examples": 50000,
      "train_loss": 0.010189418796002865
    },
    {
      "epoch": 122,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06240494176745415,
        "mean_feedback_forward_cosine": 0.9994854992086237,
        "mean_feedback_forward_relative_error": 0.030661224785514852,
        "min_feedback_forward_cosine": 0.9980674386024475
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47702,
      "train_examples": 50000,
      "train_loss": 0.009600654789209366
    },
    {
      "epoch": 123,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0623897947371006,
        "mean_feedback_forward_cosine": 0.9994856108318676,
        "mean_feedback_forward_relative_error": 0.030657286281612788,
        "min_feedback_forward_cosine": 0.9980683326721191
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48093,
      "train_examples": 50000,
      "train_loss": 0.009567934661507606
    },
    {
      "epoch": 124,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062374573200941086,
        "mean_feedback_forward_cosine": 0.9994857527992942,
        "mean_feedback_forward_relative_error": 0.03065339655361392,
        "min_feedback_forward_cosine": 0.998069167137146
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48484,
      "train_examples": 50000,
      "train_loss": 0.009035898718833923
    },
    {
      "epoch": 125,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06235916540026665,
        "mean_feedback_forward_cosine": 0.9994858785109086,
        "mean_feedback_forward_relative_error": 0.030649371428245847,
        "min_feedback_forward_cosine": 0.998070240020752
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48875,
      "train_examples": 50000,
      "train_loss": 0.00903761274561286
    },
    {
      "epoch": 126,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06234424188733101,
        "mean_feedback_forward_cosine": 0.9994860139760104,
        "mean_feedback_forward_relative_error": 0.030645305807278915,
        "min_feedback_forward_cosine": 0.9980710744857788
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49266,
      "train_examples": 50000,
      "train_loss": 0.008683195413649081
    },
    {
      "epoch": 127,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06232771649956703,
        "mean_feedback_forward_cosine": 0.9994861537759955,
        "mean_feedback_forward_relative_error": 0.030641083588654344,
        "min_feedback_forward_cosine": 0.9980721473693848
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49657,
      "train_examples": 50000,
      "train_loss": 0.008440109309703112
    },
    {
      "epoch": 128,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062310799956321716,
        "mean_feedback_forward_cosine": 0.9994863054969094,
        "mean_feedback_forward_relative_error": 0.030636796067384155,
        "min_feedback_forward_cosine": 0.998073160648346
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50048,
      "train_examples": 50000,
      "train_loss": 0.008790008682310581
    },
    {
      "epoch": 129,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06229575350880623,
        "mean_feedback_forward_cosine": 0.9994864637201483,
        "mean_feedback_forward_relative_error": 0.030632674372331664,
        "min_feedback_forward_cosine": 0.998073935508728
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50439,
      "train_examples": 50000,
      "train_loss": 0.007810075041279197
    },
    {
      "epoch": 130,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062281206250190735,
        "mean_feedback_forward_cosine": 0.9994865883480418,
        "mean_feedback_forward_relative_error": 0.03062878699803894,
        "min_feedback_forward_cosine": 0.9980750679969788
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50830,
      "train_examples": 50000,
      "train_loss": 0.007518504073172808
    },
    {
      "epoch": 131,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06226494535803795,
        "mean_feedback_forward_cosine": 0.999486716227098,
        "mean_feedback_forward_relative_error": 0.030624614342708478,
        "min_feedback_forward_cosine": 0.9980760216712952
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51221,
      "train_examples": 50000,
      "train_loss": 0.007846000881716608
    },
    {
      "epoch": 132,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062249086797237396,
        "mean_feedback_forward_cosine": 0.999486861445687,
        "mean_feedback_forward_relative_error": 0.030620227356187323,
        "min_feedback_forward_cosine": 0.9980770349502563
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51612,
      "train_examples": 50000,
      "train_loss": 0.007394560450278222
    },
    {
      "epoch": 133,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06223296746611595,
        "mean_feedback_forward_cosine": 0.9994870283386924,
        "mean_feedback_forward_relative_error": 0.030616129477593033,
        "min_feedback_forward_cosine": 0.9980779886245728
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52003,
      "train_examples": 50000,
      "train_loss": 0.00689208334473893
    },
    {
      "epoch": 134,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062215469777584076,
        "mean_feedback_forward_cosine": 0.9994871757247231,
        "mean_feedback_forward_relative_error": 0.0306115483018485,
        "min_feedback_forward_cosine": 0.9980791807174683
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52394,
      "train_examples": 50000,
      "train_loss": 0.00785630393801257
    },
    {
      "epoch": 135,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062198810279369354,
        "mean_feedback_forward_cosine": 0.9994873187758706,
        "mean_feedback_forward_relative_error": 0.030607500587674703,
        "min_feedback_forward_cosine": 0.9980801343917847
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52785,
      "train_examples": 50000,
      "train_loss": 0.007258220757246018
    },
    {
      "epoch": 136,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06218484044075012,
        "mean_feedback_forward_cosine": 0.9994874531572515,
        "mean_feedback_forward_relative_error": 0.030603864069350742,
        "min_feedback_forward_cosine": 0.9980810880661011
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 53176,
      "train_examples": 50000,
      "train_loss": 0.0061892494452372195
    },
    {
      "epoch": 137,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06217186897993088,
        "mean_feedback_forward_cosine": 0.9994875712828203,
        "mean_feedback_forward_relative_error": 0.030600314316424456,
        "min_feedback_forward_cosine": 0.9980817437171936
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 53567,
      "train_examples": 50000,
      "train_loss": 0.0059076879094541075
    },
    {
      "epoch": 138,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06215699762105942,
        "mean_feedback_forward_cosine": 0.9994876991618763,
        "mean_feedback_forward_relative_error": 0.03059661764312874,
        "min_feedback_forward_cosine": 0.99808269739151
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 53958,
      "train_examples": 50000,
      "train_loss": 0.005383817988932132
    },
    {
      "epoch": 139,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062138814479112625,
        "mean_feedback_forward_cosine": 0.9994878508827904,
        "mean_feedback_forward_relative_error": 0.030592034469273957,
        "min_feedback_forward_cosine": 0.9980838894844055
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 54349,
      "train_examples": 50000,
      "train_loss": 0.006710900447592139
    },
    {
      "epoch": 140,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06212152913212776,
        "mean_feedback_forward_cosine": 0.9994880058548667,
        "mean_feedback_forward_relative_error": 0.03058764510364695,
        "min_feedback_forward_cosine": 0.9980849623680115
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 54740,
      "train_examples": 50000,
      "train_loss": 0.006575945821553469
    },
    {
      "epoch": 141,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06210213527083397,
        "mean_feedback_forward_cosine": 0.9994881868362426,
        "mean_feedback_forward_relative_error": 0.030582719304683533,
        "min_feedback_forward_cosine": 0.998086154460907
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 55131,
      "train_examples": 50000,
      "train_loss": 0.006847478371560574
    },
    {
      "epoch": 142,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0620831735432148,
        "mean_feedback_forward_cosine": 0.9994883591478522,
        "mean_feedback_forward_relative_error": 0.030578120323744688,
        "min_feedback_forward_cosine": 0.9980872869491577
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 55522,
      "train_examples": 50000,
      "train_loss": 0.00634564375422895
    },
    {
      "epoch": 143,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06206458806991577,
        "mean_feedback_forward_cosine": 0.9994885032827204,
        "mean_feedback_forward_relative_error": 0.0305734507993541,
        "min_feedback_forward_cosine": 0.9980883598327637
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 55913,
      "train_examples": 50000,
      "train_loss": 0.006293262675926089
    },
    {
      "epoch": 144,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.062043607234954834,
        "mean_feedback_forward_cosine": 0.9994886961850253,
        "mean_feedback_forward_relative_error": 0.03056816380809654,
        "min_feedback_forward_cosine": 0.9980897307395935
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 56304,
      "train_examples": 50000,
      "train_loss": 0.006856801314167678
    },
    {
      "epoch": 145,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06202723830938339,
        "mean_feedback_forward_cosine": 0.9994888533245433,
        "mean_feedback_forward_relative_error": 0.030563697379759766,
        "min_feedback_forward_cosine": 0.9980907440185547
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 56695,
      "train_examples": 50000,
      "train_loss": 0.005869663535952568
    },
    {
      "epoch": 146,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06201047822833061,
        "mean_feedback_forward_cosine": 0.9994889909570868,
        "mean_feedback_forward_relative_error": 0.030559722884473475,
        "min_feedback_forward_cosine": 0.9980918169021606
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 57086,
      "train_examples": 50000,
      "train_loss": 0.005216972618214786
    },
    {
      "epoch": 147,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06199338287115097,
        "mean_feedback_forward_cosine": 0.9994891394268383,
        "mean_feedback_forward_relative_error": 0.030555458688600497,
        "min_feedback_forward_cosine": 0.9980928301811218
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 57477,
      "train_examples": 50000,
      "train_loss": 0.005385983447609469
    },
    {
      "epoch": 148,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06197473779320717,
        "mean_feedback_forward_cosine": 0.999489304152402,
        "mean_feedback_forward_relative_error": 0.03055039835585789,
        "min_feedback_forward_cosine": 0.998093843460083
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 57868,
      "train_examples": 50000,
      "train_loss": 0.005695979114994407
    },
    {
      "epoch": 149,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06195850670337677,
        "mean_feedback_forward_cosine": 0.999489455873316,
        "mean_feedback_forward_relative_error": 0.030546232837844977,
        "min_feedback_forward_cosine": 0.9980949759483337
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 58259,
      "train_examples": 50000,
      "train_loss": 0.005280598983112722
    },
    {
      "epoch": 150,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061936236917972565,
        "mean_feedback_forward_cosine": 0.9994896509430625,
        "mean_feedback_forward_relative_error": 0.030540636054832825,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 58650,
      "train_examples": 50000,
      "train_loss": 0.00612729561381042
    },
    {
      "epoch": 151,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193609535694122,
        "mean_feedback_forward_cosine": 0.9994896455244584,
        "mean_feedback_forward_relative_error": 0.03054057719524611,
        "min_feedback_forward_cosine": 0.998096227645874
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59041,
      "train_examples": 50000,
      "train_loss": 0.004530388177447021
    },
    {
      "epoch": 152,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193596124649048,
        "mean_feedback_forward_cosine": 0.999489659612829,
        "mean_feedback_forward_relative_error": 0.030540518132461742,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59432,
      "train_examples": 50000,
      "train_loss": 0.004308553957641125
    },
    {
      "epoch": 153,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061935849487781525,
        "mean_feedback_forward_cosine": 0.9994896574453874,
        "mean_feedback_forward_relative_error": 0.030540465174073522,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59823,
      "train_examples": 50000,
      "train_loss": 0.004125927190994844
    },
    {
      "epoch": 154,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061935700476169586,
        "mean_feedback_forward_cosine": 0.9994896509430625,
        "mean_feedback_forward_relative_error": 0.030540410141375932,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60214,
      "train_examples": 50000,
      "train_loss": 0.00451846117965877
    },
    {
      "epoch": 155,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193555146455765,
        "mean_feedback_forward_cosine": 0.9994896617802707,
        "mean_feedback_forward_relative_error": 0.03054035593840209,
        "min_feedback_forward_cosine": 0.9980962872505188
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60605,
      "train_examples": 50000,
      "train_loss": 0.004326087524071336
    },
    {
      "epoch": 156,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061935409903526306,
        "mean_feedback_forward_cosine": 0.9994896531105042,
        "mean_feedback_forward_relative_error": 0.030540305858647282,
        "min_feedback_forward_cosine": 0.998096227645874
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60996,
      "train_examples": 50000,
      "train_loss": 0.004024175264686346
    },
    {
      "epoch": 157,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193528696894646,
        "mean_feedback_forward_cosine": 0.9994896563616666,
        "mean_feedback_forward_relative_error": 0.0305402558804913,
        "min_feedback_forward_cosine": 0.998096227645874
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61387,
      "train_examples": 50000,
      "train_loss": 0.003828567948546261
    },
    {
      "epoch": 158,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193516030907631,
        "mean_feedback_forward_cosine": 0.9994896617802707,
        "mean_feedback_forward_relative_error": 0.03054020767184821,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61778,
      "train_examples": 50000,
      "train_loss": 0.004268716067522764
    },
    {
      "epoch": 159,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193503737449646,
        "mean_feedback_forward_cosine": 0.9994896693663163,
        "mean_feedback_forward_relative_error": 0.030540157355029473,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62169,
      "train_examples": 50000,
      "train_loss": 0.0038033735569939016
    },
    {
      "epoch": 160,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193489581346512,
        "mean_feedback_forward_cosine": 0.9994896802035245,
        "mean_feedback_forward_relative_error": 0.03054010575975884,
        "min_feedback_forward_cosine": 0.9980962872505188
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62560,
      "train_examples": 50000,
      "train_loss": 0.003953896807860583
    },
    {
      "epoch": 161,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193477660417557,
        "mean_feedback_forward_cosine": 0.9994896704500372,
        "mean_feedback_forward_relative_error": 0.030540055095810783,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62951,
      "train_examples": 50000,
      "train_loss": 0.0038096044862642882
    },
    {
      "epoch": 162,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061934683471918106,
        "mean_feedback_forward_cosine": 0.9994896758686412,
        "mean_feedback_forward_relative_error": 0.030540014862675557,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63342,
      "train_examples": 50000,
      "train_loss": 0.0033296207774430515
    },
    {
      "epoch": 163,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061934567987918854,
        "mean_feedback_forward_cosine": 0.9994896531105042,
        "mean_feedback_forward_relative_error": 0.03053997020152482,
        "min_feedback_forward_cosine": 0.9980962872505188
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63733,
      "train_examples": 50000,
      "train_loss": 0.0036760004641488197
    },
    {
      "epoch": 164,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0619344562292099,
        "mean_feedback_forward_cosine": 0.9994896693663163,
        "mean_feedback_forward_relative_error": 0.0305399274114858,
        "min_feedback_forward_cosine": 0.998096227645874
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64124,
      "train_examples": 50000,
      "train_loss": 0.003656223715506494
    },
    {
      "epoch": 165,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193436309695244,
        "mean_feedback_forward_cosine": 0.9994896780360829,
        "mean_feedback_forward_relative_error": 0.030539883520792833,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64515,
      "train_examples": 50000,
      "train_loss": 0.003014061942361295
    },
    {
      "epoch": 166,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061934247612953186,
        "mean_feedback_forward_cosine": 0.9994896661151539,
        "mean_feedback_forward_relative_error": 0.03053983548148112,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64906,
      "train_examples": 50000,
      "train_loss": 0.003544073878750205
    },
    {
      "epoch": 167,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193413957953453,
        "mean_feedback_forward_cosine": 0.9994896639477123,
        "mean_feedback_forward_relative_error": 0.030539789008484645,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65297,
      "train_examples": 50000,
      "train_loss": 0.003221760457754135
    },
    {
      "epoch": 168,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193403899669647,
        "mean_feedback_forward_cosine": 0.9994896747849205,
        "mean_feedback_forward_relative_error": 0.03053974864835089,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65688,
      "train_examples": 50000,
      "train_loss": 0.003004233146267943
    },
    {
      "epoch": 169,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193394958972931,
        "mean_feedback_forward_cosine": 0.9994896628639914,
        "mean_feedback_forward_relative_error": 0.030539709736000408,
        "min_feedback_forward_cosine": 0.9980964064598083
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66079,
      "train_examples": 50000,
      "train_loss": 0.0030014431909099223
    },
    {
      "epoch": 170,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061933841556310654,
        "mean_feedback_forward_cosine": 0.9994896726174788,
        "mean_feedback_forward_relative_error": 0.030539668173613872,
        "min_feedback_forward_cosine": 0.9980962872505188
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66470,
      "train_examples": 50000,
      "train_loss": 0.0033193327287212014
    },
    {
      "epoch": 171,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0619337260723114,
        "mean_feedback_forward_cosine": 0.9994896682825956,
        "mean_feedback_forward_relative_error": 0.030539624198255212,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66861,
      "train_examples": 50000,
      "train_loss": 0.0032627053089253604
    },
    {
      "epoch": 172,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193362548947334,
        "mean_feedback_forward_cosine": 0.9994896726174788,
        "mean_feedback_forward_relative_error": 0.030539582204073667,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67252,
      "train_examples": 50000,
      "train_loss": 0.0028701702140644194
    },
    {
      "epoch": 173,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193353980779648,
        "mean_feedback_forward_cosine": 0.9994896563616666,
        "mean_feedback_forward_relative_error": 0.03053954222493551,
        "min_feedback_forward_cosine": 0.9980962872505188
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67643,
      "train_examples": 50000,
      "train_loss": 0.0028575154671818016
    },
    {
      "epoch": 174,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193346157670021,
        "mean_feedback_forward_cosine": 0.9994896780360829,
        "mean_feedback_forward_relative_error": 0.030539504176175052,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68034,
      "train_examples": 50000,
      "train_loss": 0.0030732198222354055
    },
    {
      "epoch": 175,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061933357268571854,
        "mean_feedback_forward_cosine": 0.9994896802035245,
        "mean_feedback_forward_relative_error": 0.030539463240314615,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68425,
      "train_examples": 50000,
      "train_loss": 0.0028843802531808614
    },
    {
      "epoch": 176,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061933282762765884,
        "mean_feedback_forward_cosine": 0.9994896758686412,
        "mean_feedback_forward_relative_error": 0.03053942309184508,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68816,
      "train_examples": 50000,
      "train_loss": 0.002822302868962288
    },
    {
      "epoch": 177,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193319708108902,
        "mean_feedback_forward_cosine": 0.9994896747849205,
        "mean_feedback_forward_relative_error": 0.03053938172418963,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69207,
      "train_examples": 50000,
      "train_loss": 0.0028555231500044464
    },
    {
      "epoch": 178,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061933111399412155,
        "mean_feedback_forward_cosine": 0.9994896856221286,
        "mean_feedback_forward_relative_error": 0.030539340161803093,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69598,
      "train_examples": 50000,
      "train_loss": 0.002892176869865507
    },
    {
      "epoch": 179,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193302571773529,
        "mean_feedback_forward_cosine": 0.9994896758686412,
        "mean_feedback_forward_relative_error": 0.030539298768747936,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69989,
      "train_examples": 50000,
      "train_loss": 0.0029899260829761624
    },
    {
      "epoch": 180,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193292886018753,
        "mean_feedback_forward_cosine": 0.9994896780360829,
        "mean_feedback_forward_relative_error": 0.030539256740700117,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70380,
      "train_examples": 50000,
      "train_loss": 0.002768810143917799
    },
    {
      "epoch": 181,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193281337618828,
        "mean_feedback_forward_cosine": 0.9994896802035245,
        "mean_feedback_forward_relative_error": 0.03053921466185288,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70771,
      "train_examples": 50000,
      "train_loss": 0.003306742629110813
    },
    {
      "epoch": 182,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061932727694511414,
        "mean_feedback_forward_cosine": 0.9994896932081743,
        "mean_feedback_forward_relative_error": 0.030539175173775718,
        "min_feedback_forward_cosine": 0.9980965256690979
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 71162,
      "train_examples": 50000,
      "train_loss": 0.0029128653236106038
    },
    {
      "epoch": 183,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061932627111673355,
        "mean_feedback_forward_cosine": 0.9994896953756159,
        "mean_feedback_forward_relative_error": 0.030539133357392115,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 71553,
      "train_examples": 50000,
      "train_loss": 0.002987928712326102
    },
    {
      "epoch": 184,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193254888057709,
        "mean_feedback_forward_cosine": 0.9994896812872454,
        "mean_feedback_forward_relative_error": 0.03053909654475071,
        "min_feedback_forward_cosine": 0.9980963468551636
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 71944,
      "train_examples": 50000,
      "train_loss": 0.0026746951858513056
    },
    {
      "epoch": 185,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193244829773903,
        "mean_feedback_forward_cosine": 0.9994896877895701,
        "mean_feedback_forward_relative_error": 0.030539051570336928,
        "min_feedback_forward_cosine": 0.9980965852737427
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 72335,
      "train_examples": 50000,
      "train_loss": 0.0030044303303956984
    },
    {
      "epoch": 186,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193232908844948,
        "mean_feedback_forward_cosine": 0.9994896932081743,
        "mean_feedback_forward_relative_error": 0.03053900718858296,
        "min_feedback_forward_cosine": 0.9980965852737427
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 72726,
      "train_examples": 50000,
      "train_loss": 0.0030119929092749954
    },
    {
      "epoch": 187,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193223595619202,
        "mean_feedback_forward_cosine": 0.99948970079422,
        "mean_feedback_forward_relative_error": 0.030538960571654818,
        "min_feedback_forward_cosine": 0.9980964064598083
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 73117,
      "train_examples": 50000,
      "train_loss": 0.003049773293584585
    },
    {
      "epoch": 188,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193213537335396,
        "mean_feedback_forward_cosine": 0.9994896953756159,
        "mean_feedback_forward_relative_error": 0.03053891994059086,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 73508,
      "train_examples": 50000,
      "train_loss": 0.0029004252321086824
    },
    {
      "epoch": 189,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0619320422410965,
        "mean_feedback_forward_cosine": 0.9994896791198037,
        "mean_feedback_forward_relative_error": 0.030538882730020717,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 73899,
      "train_examples": 50000,
      "train_loss": 0.0028327848418429496
    },
    {
      "epoch": 190,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193194538354874,
        "mean_feedback_forward_cosine": 0.9994896986267784,
        "mean_feedback_forward_relative_error": 0.030538840193978765,
        "min_feedback_forward_cosine": 0.9980965852737427
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 74290,
      "train_examples": 50000,
      "train_loss": 0.0027145167157426476
    },
    {
      "epoch": 191,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193183735013008,
        "mean_feedback_forward_cosine": 0.9994897051291032,
        "mean_feedback_forward_relative_error": 0.030538800832900136,
        "min_feedback_forward_cosine": 0.9980964064598083
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 74681,
      "train_examples": 50000,
      "train_loss": 0.0029637707200273872
    },
    {
      "epoch": 192,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061931759119033813,
        "mean_feedback_forward_cosine": 0.9994896899570118,
        "mean_feedback_forward_relative_error": 0.030538763639263132,
        "min_feedback_forward_cosine": 0.9980966448783875
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 75072,
      "train_examples": 50000,
      "train_loss": 0.002693286809436977
    },
    {
      "epoch": 193,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061931658536195755,
        "mean_feedback_forward_cosine": 0.9994897029616616,
        "mean_feedback_forward_relative_error": 0.030538724193518812,
        "min_feedback_forward_cosine": 0.9980966448783875
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 75463,
      "train_examples": 50000,
      "train_loss": 0.002560252621918917
    },
    {
      "epoch": 194,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0619315542280674,
        "mean_feedback_forward_cosine": 0.9994897051291032,
        "mean_feedback_forward_relative_error": 0.03053867809305137,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 75854,
      "train_examples": 50000,
      "train_loss": 0.002967231141254306
    },
    {
      "epoch": 195,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.061931461095809937,
        "mean_feedback_forward_cosine": 0.9994897094639865,
        "mean_feedback_forward_relative_error": 0.030538642626594413,
        "min_feedback_forward_cosine": 0.9980966448783875
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 76245,
      "train_examples": 50000,
      "train_loss": 0.002552174325212836
    },
    {
      "epoch": 196,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193136051297188,
        "mean_feedback_forward_cosine": 0.9994896986267784,
        "mean_feedback_forward_relative_error": 0.030538603087717837,
        "min_feedback_forward_cosine": 0.9980965852737427
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 76636,
      "train_examples": 50000,
      "train_loss": 0.0028820837849471716
    },
    {
      "epoch": 197,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193125247955322,
        "mean_feedback_forward_cosine": 0.9994897051291032,
        "mean_feedback_forward_relative_error": 0.030538562515919857,
        "min_feedback_forward_cosine": 0.9980964660644531
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 77027,
      "train_examples": 50000,
      "train_loss": 0.002910595020055771
    },
    {
      "epoch": 198,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193114072084427,
        "mean_feedback_forward_cosine": 0.9994897094639865,
        "mean_feedback_forward_relative_error": 0.03053851816803217,
        "min_feedback_forward_cosine": 0.9980965852737427
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 77418,
      "train_examples": 50000,
      "train_loss": 0.00310367449041456
    },
    {
      "epoch": 199,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193104013800621,
        "mean_feedback_forward_cosine": 0.9994897072965448,
        "mean_feedback_forward_relative_error": 0.03053847612305121,
        "min_feedback_forward_cosine": 0.9980965256690979
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 77809,
      "train_examples": 50000,
      "train_loss": 0.0028941941729560493
    },
    {
      "epoch": 200,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06193094700574875,
        "mean_feedback_forward_cosine": 0.9994896823709661,
        "mean_feedback_forward_relative_error": 0.03053843260488727,
        "min_feedback_forward_cosine": 0.9980965256690979
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 78200,
      "train_examples": 50000,
      "train_loss": 0.0028094120770692827
    }
  ],
  "evaluation_protocol": {
    "test_evaluations": 1,
    "test_used_for_selection": false,
    "validation_evaluations": 0
  },
  "final": {
    "accuracy": 0.9287,
    "epoch": 200,
    "evaluation_split": "test",
    "finite": true,
    "loss": 0.3952233123779297,
    "step": 78200
  },
  "hardware": {
    "cuda_device_name": "NVIDIA GeForce GTX 1080",
    "cuda_visible_devices": "5",
    "device": "cuda",
    "device_total_memory_bytes": 8507949056,
    "peak_memory_allocated_bytes": 3318545920,
    "peak_memory_reserved_bytes": 4754243584,
    "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": 10,
    "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": 2.0534486770629883,
    "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": 18251.1642100811,
    "train_wall_s": 18245.253133296967
  },
  "work": {
    "apical_macs_per_example": 125043328,
    "causal_scalar_observations": 0,
    "components": {
      "apical_projection_macs": 1250433280000000,
      "apical_regression_macs": 0,
      "bp_reverse_macs_estimate": 0,
      "kp_reciprocal_correlation_macs": 1250433280000000,
      "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": 5010580480000000
  }
}